Dissertations / Theses on the topic 'Dense networks'

The traffic demands predicted by 2030 are up to 10,000 times greater than in 2010 and end-service users will need to support 100 Mbps. One of the key developments that will provide this demand is the deployment of very dense and multi layered networks.

Neural networks can perform an incredible array of complex tasks, but successfully training a network is difficult because it requires us to minimize a function about which we know very little. In practice, developing a good model requires both intuition and a lot of guess-and-check. In this dissertation, we study a type of fully-connected neural network that improves on standard rectifier networks while retaining their useful properties. We then examine this type of network and its loss function from a probabilistic perspective. This analysis leads to a new rule for parameter initialization and a new method for predicting effective learning rates for gradient descent. Experiments confirm that the theory behind these developments translates well into practice.

Mobile data traffic is expected to increase substantially in the coming years, with data rates 1000 times higher by 2020, having media and content as the main drivers together with a plethora of new end-user services that will challenge existing networks. Concepts and visions associated with the ICT evolution like the network society, 50 billion connected devices, Industrial Internet, Tactile Internet, etc., exemplifies the range of new services that the networks will have to handle. These new services impose extreme requirement to the network like high capacity, low latency, reliability, security, seamless connectivity, etc. In order to face these challenges, the whole end-to-end network has to evolve and adapt, pushing for advances in different areas, such as transport, cloud, core, and radio access networks. This work investigates the impact of envisioned 2020 society scenarios on transport links for mobile backhaul, emphasizing the need for an integrated and flexible/adaptive network as the way to meet the 2020 networks demands. The evolution of heterogeneous networks and ultra-dense network deployments shall also comprise the introduction of adaptive network features, such as dynamic network resource allocation, automatic integration of access nodes, etc. In order to achieve such self-management features in mobile networks, new mechanisms have to be investigated for an integrated backhaul management. First, this thesis performs a feasibility study on the mobile backhaul dimensioning for 2020 5G wireless ultra-dense networks scenarios, aiming to analyze the gap in capacity demand between 4G and 5G networks. Secondly, the concept of an integrated backhaul management is analyzed as a combination of node attachment procedures, in the context of moving networks. In addition, the dynamic network resource allocation concept, based on DWDM-centric transport architecture, was explored for 5G scenarios assuming traffic variation both in time and between different geographical areas. Finally, a short view on techno-economics and network deployments in the 2020 time frame is provided.

In this project, we investigate the performance of Raptor codes as candidatesfor channel coding for the wireless communication between access nodes.Very high data-rates are used, and processing uses more resources than transmission.Therefore, we need fast encoding and decoding algorithms for thechannel coding. Raptor codes have linear encoding and decoding times, andcan have very small overhead if they are properly designed. Hence, they arepossible candidates. We have implemented an encoding and decoding algorithm for Raptorcodes, as well as an environment for simulation. The system requirementsare expressed in terms of delay between the beginning of a transmission andthe successful decoding, and storage required during the transmission and processing.We have evaluated the performance of Raptor codes in terms of delayand storage as a function of system design parameters, in particular the numberof nodes in the network, and the size of the packets. We show that if thesize of the packets is properly chosen, Raptor codes can be useful for the application,and we explain the method for choosing the size of the packets. Wealso provide a way to calculate the delay and the storage for a given systemconfiguration, in order for example to determinate the larger number of nodesor the larger of users such that the delay and the storage are acceptable.

In the continuum modeling of traffic networks, a macroscopic cost-flow function (MCF) and macroscopic fundamental diagram (MFD) can be used to represent the fundamental relationships between traffic quantities such as speed, flow, and density. The MCF governs the steady-state cost-flow relationship, whereas the MFD represents the instantaneous inter-relationship between speed, flow, and density of traffic streams. This thesis explores the influence of network topologies on the MCF and MFD. The Hong Kong road system is divided into unit-sized road networks with various physical characteristics for which the network structure and signal timings are reserved. By universally scaling the origin-destination (OD) matrices of the morning peak, traffic conditions ranging from free-flow to congestion are created for microscopic simulation. From the simulation results, an MCF that relates the average journey time and the number of vehicles traveling through the network in one hour and an MFD that relates space to the mean speed and average density aggregated across 300s intervals are derived. The MCF and MFD are calibrated with mathematical models for each network. The density of roads, junctions, and signal junctions all influence the value of the macroscopic parameters in the MCF and MFD, and predictive equations are constructed that relate the macroscopic parameters to the network topological characteristics. Based on the fitting performance of the mathematical models, recommendations are made for selecting MCF and MFD models for continuum modeling.
published_or_final_version
Civil Engineering
Master
Master of Philosophy

Wireless communication systems have exploded in popularity over the past few decades. Due to their popularity, the demand for higher data rates by the users, and the high cost of wireless spectrum, wireless providers are actively seeking ways to improve the spectral efficiency of their networks. One promising technique to improve spectral efficiency is to equip the wireless devices with multiple antennas. If both the transmitter and receiver of a link are equipped with multiple antennas, they form a multiple-input multiple-output (MIMO) link. The multiple antennas at the nodes provide degrees-of-freedom that can be used for either sending multiple streams of data simultaneously (a technique known as spatial multiplexing), or for suppressing interference through linear combining, but not both. Due to this trade-off, careful allocation of how many streams each link should carry is important to ensure that each node has enough degrees-of-freedom available to suppress the interference and support its desired streams. How the streams are sent and received and how interference is suppressed is ultimately determined by the beamforming weights at the transmitters and the combining weights at the receivers. Determining these weights is, however, made difficult by their inherent interdependency. Our focus is on unplanned and/or dense single-hop networks, such as WLANs and femtocells, where each single-hop network is composed of an access point serving several associated clients. The objective of this research is to design algorithms for maximizing the performance of dense single-hop wireless networks of MIMO links. We address the problems of determining which links to schedule together at each time slot, how many streams to allocate to each link (if any), and the beamforming and combining weights that support those streams. This dissertation describes four key contributions as follows: - We classify any interference suppression technique as either unilateral interference suppression or bilateral interference suppression. We show that a simple bilateral interference suppression approach outperforms all known unilateral interference suppression approaches, even after searching for the best unilateral solution. - We propose an algorithm based on bilateral interference suppression whose goal is to maximize the sum rate of a set of interfering MIMO links by jointly optimizing which subset of transmitters should transmit, the number of streams for each transmitter (if any), and the beamforming and combining weights that support those streams. - We propose a framework for optimizing dense single-hop wireless networks. The framework implements techniques to address several practical issues that arise when implementing interference suppression, such as the overhead of performing channel measurements and communicating channel state information, the overhead of computing the beamforming and combining weights, and the overhead of cooperation between the access points. - We derive the optimal scheduler that maximizes the sum rate subject to proportional fairness. Simulations in ns-3 show that the framework, using the optimal scheduler, increases the proportionally fair aggregate goodput by up to 165% as compared to the aggregate goodput of 802.11n for the case of four interfering single-hop wireless networks with two clients each.

A new communication paradigm is foreseen for beyond 2020 society, due to the emergence of new broadband services and the Internet of Things era. The set of requirements imposed by these new applications is large and diverse, aiming to provide a ubiquitous broadband connectivity. Research community has been working in the last decade towards the definition of the 5G mobile wireless networks that will provide the proper mechanisms to reach these challenging requirements. In this framework, three key research directions have been identified for the improvement of capacity in 5G: the increase of the spectral efficiency by means of, for example, the use of massive MIMO technology, the use of larger amounts of spectrum by utilizing the millimeter wave band, and the network densification by deploying more base stations per unit area. This dissertation addresses densification as the main enabler for the broadband and massive connectivity required in future 5G networks. To this aim, this Thesis focuses on the study of the UDN. In particular, a set of technology enablers that can lead UDN to achieve their maximum efficiency and performance are investigated, namely, the use of higher frequency bands for the benefit of larger bandwidths, the use of massive MIMO with distributed antenna systems, and the use of distributed radio resource management techniques for the inter-cell interference coordination. Firstly, this Thesis analyzes whether there exists a fundamental performance limit related with densification in cellular networks. To this end, the UDN performance is evaluated by means of an analytical model consisting of a 1-dimensional network deployment with equally spaced BS. The inter-BS distance is decreased until reaching the limit of densification when this distance approaches 0. The achievable rates in networks with different inter-BS distances are analyzed for several levels of transmission power availability, and for various types of cooperation among cells. Moreover, UDN performance is studied in conjunction with the use of a massive number of antennas and larger amounts of spectrum. In particular, the performance of hybrid beamforming and precoding MIMO schemes are assessed in both indoor and outdoor scenarios with multiple cells and users, working in the mmW frequency band. On the one hand, beamforming schemes using the full-connected hybrid architecture are analyzed in BS with limited number of RF chains, identifying the strengths and weaknesses of these schemes in a dense-urban scenario. On the other hand, the performance of different indoor deployment strategies using HP in the mmW band is evaluated, focusing on the use of DAS. More specifically, a DHP suitable for DAS is proposed, comparing its performance with that of HP in other indoor deployment strategies. Lastly, the presence of practical limitations and hardware impairments in the use of hybrid architectures is also investigated. Finally, the investigation of UDN is completed with the study of their main limitation, which is the increasing inter-cell interference in the network. In order to tackle this problem, an eICIC scheduling algorithm based on resource partitioning techniques is proposed. Its performance is evaluated and compared to other scheduling algorithms under several degrees of network densification. After the completion of this study, the potential of UDN to reach the capacity requirements of 5G networks is confirmed. Nevertheless, without the use of larger portions of spectrum, a proper interference management and the use of a massive number of antennas, densification could turn into a serious problem for mobile operators. Performance evaluation results show large system capacity gains with the use of massive MIMO techniques in UDN, and even greater when the antennas are distributed. Furthermore, the application of ICIC techniques reveals that, besides the increase in system capacity, it brings significant energy savings to UDNs.
A partir del año 2020 se prevé que un nuevo paradigma de comunicación surja en la sociedad, debido a la aparición de nuevos servicios y la era del Internet de las cosas. El conjunto de requisitos impuesto por estas nuevas aplicaciones es muy amplio y diverso, y tiene como principal objetivo proporcionar conectividad de banda ancha y universal. En las últimas décadas, la comunidad científica ha estado trabajando en la definición de la 5G de redes móviles que brindará los mecanismos necesarios para garantizar estos requisitos. En este marco, se han identificado tres mecanismos clave para conseguir el necesario incremento de capacidad de la red: el aumento de la eficiencia espectral a través de, por ejemplo, el uso de tecnologías MIMO masivas, la utilización de mayores porciones del espectro en frecuencia y la densificación de la red mediante el despliegue de más estaciones base por área. Esta Tesis doctoral aborda la densificación como el principal mecanismo que permitirá la conectividad de banda ancha y universal requerida en la 5G, centrándose en el estudio de las Redes Ultra Densas o UDNs. En concreto, se analiza el conjunto de tecnologías habilitantes que pueden llevar a las UDNs a obtener su máxima eficiencia y prestaciones, incluyendo el uso de altas frecuencias para el aprovechamiento de mayores anchos de banda, la utilización de MIMO masivo con sistemas de antenas distribuidas y el uso de técnicas de reparto de recursos distribuidas para la coordinación de interferencias. En primer lugar, se analiza si existe un límite fundamental en la mejora de las prestaciones en relación a la densificación. Con este fin, las prestaciones de las UDNs se evalúan utilizando un modelo analítico de red unidimensional con BSs equiespaciadas, en el que la distancia entre BSs se disminuye hasta alcanzar el límite de densificación cuando ésta se aproxima a 0. Las tasas alcanzables en redes con distintas distancias entre BSs son analizadas, considerando distintos niveles de potencia disponible en la red y varios grados de cooperación entre celdas. Además, el comportamiento de las UDNs se estudia junto al uso masivo de antenas y la utilización de anchos de banda mayores. Más concretamente, las prestaciones de ciertas técnicas híbridas MIMO de precodificación y beamforming se examinan en la banda milimétrica. Por una parte, se analizan esquemas de beamforming en BSs con arquitectura híbrida en función de la disponibilidad de cadenas de radiofrecuencia en escenarios exteriores. Por otra parte, se evalúan las prestaciones de ciertos esquemas de precodificación híbrida en escenarios interiores, utilizando distintos despliegues y centrando la atención en los sistemas de antenas distribuidos o DAS. Además, se propone un algoritmo de precodificación híbrida específico para DAS, y se evalúan y comparan sus prestaciones con las de otros algoritmos de precodificación utilizados. Por último, se investiga el impacto en las prestaciones de ciertas limitaciones prácticas y deficiencias introducidas por el uso de dispositivos no ideales. Finalmente, el estudio de las UDNs se completa con el análisis de su principal limitación, el nivel creciente de interferencia en la red. Para ello, se propone un algoritmo de control de interferencias basado en la partición de recursos. Sus prestaciones son evaluadas y comparadas con las de otras técnicas de asignación de recursos. Tras este estudio, se puede afirmar que las UDNs tienen gran potencial para la consecución de los requisitos de la 5G. Sin embargo, sin el uso conjunto de mayores porciones del espectro, adecuadas técnicas de control de la interferencia y el uso masivo de antenas, las UDNs pueden convertirse en serios obstáculos para los operadores móviles. Los resultados de la evaluación de prestaciones de estas tecnologías confirman el gran aumento de la capacidad de las redes mediante el uso masivo de antenas y la introducción de mecanismos de I
A partir de l'any 2020 es preveu un nou paradigma de comunicació en la societat, degut a l'aparició de nous serveis i la era de la Internet de les coses. El conjunt de requeriments imposat per aquestes noves aplicacions és ampli i divers, i té com a principal objectiu proporcionar connectivitat universal i de banda ampla. En les últimes dècades, la comunitat científica ha estat treballant en la definició de la 5G, que proveirà els mecanismes necessaris per a garantir aquests exigents requeriments. En aquest marc, s'han identificat tres mecanismes claus per a aconseguir l'increment necessari en la capacitat: l'augment de l'eficiència espectral a través de, per exemple, l'ús de tecnologies MIMO massives, la utilització de majors porcions de l'espectre i la densificació mitjançant el desplegament de més estacions base per àrea. Aquesta Tesi aborda la densificació com a principal mecanisme que permetrà la connectivitat de banda ampla i universal requerida en la 5G, centrant-se en l' estudi de les xarxes ultra denses (UDNs). Concretament, el conjunt de tecnologies que poden dur a les UDNs a la seua màxima eficiència i prestacions és analitzat, incloent l'ús d'altes freqüències per a l'aprofitament de majors amplàries de banda, la utilització de MIMO massiu amb sistemes d'antenes distribuïdes i l'ús de tècniques distribuïdes de repartiment de recursos per a la coordinació de la interferència. En primer lloc, aquesta Tesi analitza si existeix un límit fonamental en les prestacions en relació a la densificació. Per això, les prestacions de les UDNs s'avaluen utilitzant un model analític unidimensional amb estacions base equidistants, en les quals la distància entre estacions base es redueix fins assolir el límit de densificació quan aquesta distància s'aproxima a 0. Les taxes assolibles en xarxes amb diferents distàncies entre estacions base s'analitzen considerant diferents nivells de potència i varis graus de cooperació entre cel·les. A més, el comportament de les UDNs s'estudia conjuntament amb l'ús massiu d'antenes i la utilització de majors amplàries de banda. Més concretament, les prestacions de certes tècniques híbrides MIMO de precodificació i beamforming s'examinen en la banda mil·limètrica. D'una banda, els esquemes de beamforming aplicats a estacions base amb arquitectures híbrides és analitzat amb disponibilitat limitada de cadenes de radiofreqüència a un escenari urbà dens. D'altra banda, s'avaluen les prestacions de certs esquemes de precodificació híbrida en escenaris d'interior, utilitzant diferents estratègies de desplegament i centrant l'atenció en els sistemes d' antenes distribuïdes (DAS). A més, es proposa un algoritme de precodificació híbrida distribuïda per a DAS, i s'avaluen i comparen les seues prestacions amb les de altres algoritmes. Per últim, s'investiga l'impacte de les limitacions pràctiques i altres deficiències introduïdes per l'ús de dispositius no ideals en les prestacions de tots els esquemes anteriors. Finalment, l' estudi de les UDNs es completa amb l'anàlisi de la seua principal limitació, el nivell creixent d'interferència entre cel·les. Per tractar aquest problema, es proposa un algoritme de control d'interferències basat en la partició de recursos. Les prestacions de l'algoritme proposat s'avaluen i comparen amb les d'altres tècniques d'assignació de recursos. Una vegada completat aquest estudi, es pot afirmar que les UDNs tenen un gran potencial per aconseguir els ambiciosos requeriments plantejats per a la 5G. Tanmateix, sense l'ús conjunt de majors amplàries de banda, apropiades tècniques de control de la interferència i l'ús massiu d'antenes, les UDNs poden convertir-se en seriosos obstacles per als operadors mòbils. Els resultats de l'avaluació de prestacions d' aquestes tecnologies confirmen el gran augment de la capacitat de les xarxes obtingut mitjançant l'ús massiu d'antenes i la introducci
Giménez Colás, S. (2017). Ultra Dense Networks Deployment for beyond 2020 Technologies [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/86204
TESIS

The increase in demand for more network capacity has led to the evolution of wireless networks from being largely Heterogeneous (Het-Nets) to the now existing Ultra-dense (UDNs). In UDNs, small cells are densely deployed with the goal of shortening the physical distance between the base stations (BSs) and the UEs, so as to support more user equipment (UEs) at peak times while ensuring high data rates. Compared to Het-Nets, Ultra-dense networks (UDNs) have many advantages. These include, more network capacity, higher flexibility to routine configurations, and more suitability to achieve load-balancing, hence, fewer blind spots as well as lower call blocking probability. It should be noted that, in practice, due to the high density of deployed small cells in Ultra-Dense Networks, a number of issues, or rather concerns, come with this evolution from Het-Nets. Among these issues include problems with efficient radio resource management, user cell association, inter- and intra-cell interference management and, last but not least, efficient energy consumption. Some of these issues which impact the overall network efficiency are largely due to the use of obsolete algorithms, especially those whose resource allocation is based solely on received signal power (RSSP). In this paper, the focus is solely on the efficient energy management dilemma and how to optimally reduce the overall network energy consumption. Through an extensive literature review, a detailed report into the growing concern of efficient energy management in UDNs is provided in Chapter 2. The literature review report highlights the classification as well as the evolution of some of the Mobile Wireless Technologies and Mobile Wireless Networks in general. The literature review report provides reasons as to why the energy consumption issue has become a very serious concern in UltraDense networks as well as the various techniques and measures taken to mitigate this. It is shown that, due to the increasing Mobile Wireless Systems’ carbon footprint which carries serious negative environmental impact, and the general need to lower operating costs by the network operators, the management of energy consumption increases in priority. By using the architecture of a Fourth Generation Long Term Evolution (4G-LTE) UltraDense Network, the report further shows that more than 65% of the overall energy consumption is by the access network and base stations in particular. This phenomenon explains why most attention in energy efficiency management in UDNs is largely centred on reducing the energy consumption of the deployed base stations more than any other network components like the data servers or backhauling features used. Furthermore, the report also provides detailed information on the methods/techniques, their classification, implementation, as well as a critical analysis of the said implementations in literature. This study proposes a sub-optimal algorithm and Distributed Cell Resource Allocation with a Base Station On/Off scheme that aims at reducing the overall base station power consumption in UDNs, while ensuring that the overall Quality of Service (QoS) for each User Equipment (UE) as specified in its service class is met. The modeling of the system model used and hence formulation of the Network Energy Efficiency (NEE) optimization problem is done viii using stochastic geometry. The network model comprises both evolved Node B (eNB) type macro and small cells operating on different frequency bands as well as taking into account factors that impact NEE such as UE mobility, UE spatial distribution and small cells spatial distribution. The channel model takes into account signal interference from all base stations, path loss, fading, log normal shadowing, modulation and coding schemes used on each UE’s communication channels when computing throughout. The power consumption model used takes into account both static (site cooling, circuit power) and active (transmission or load based) base station power consumption. The formulation of the NEE optimization problem takes into consideration the user’s Quality-of-service (QoS), inter-cell interference, as well as each user’s spectral efficiency and coverage/success probability. The formulated NEE optimization problem is of type Nondeterministic Polynomial time (NP)-hard, due to the user-cell association. The proposed solution to the formulated optimization problem makes use of constraint relaxation to transform the NP-hard problem into a more solvable, convex and linear optimization one. This, combined with Lagrangian dual decomposition, is used to create a distributed solution. After cellassociation and resource allocation phases, the proposed solution in order to further reduce power consumption performs Cell On/Off. Then, by using the computer simulation tools/environments, the “Distributed Resource Allocation with Cell On/Off” scheme’s performance, in comparison to four other resource allocation schemes, is analysed and evaluated given a number of different network scenarios. Finally, the statistical and mathematical results generated through the simulations indicate that the proposed scheme is the closest in NEE performance to the Exhaustive Search algorithm, and hence superior to the other sub-optimal algorithms it is compared to.

In 5G systems, two radio air interfaces, evolved LTE and New Radio (NR), will coexist. By using millimeter waves, NR will provide high throughputs, but the higher frequencies will also lead to increased losses and a worse coverage. Multi-connectivity is therefore envisioned as a way to tackle these effects by connecting to multiple base stations simultaneously, allowing users to benefit from both air interfaces’ advantages. In this thesis, we investigate how multi-connectivity can be used efficiently in ultra-dense networks, a new paradigm in which the number of access nodes exceeds the number of users within the network. A framework for secondary cell association is presented and an energy efficiency’s condition is proposed. Upper and lower bounds of the network’s energy efficiency are analytically expressed. Algorithms for secondary cell selection are designed and evaluated through simulations. Multi-connectivity showed an improvement of up to 50% in reliability and and an increase of up to 20% in energy efficiency.

Thesis (Ph. D.) -- University of Maryland, College Park, 2006.
Thesis research directed by: Electrical Engineering. Title from t.p. of PDF. Includes bibliographical references. Published by UMI Dissertation Services, Ann Arbor, Mich. Also available in paper.

Because of the rapid improvement of wireless communication, the requirement for data traffic is increasing faster and faster. How to provide enough service for the increasing user amount is a problem while designing the wireless systems. Making the wireless network more density seems to be an obvious solution. But more base stations means more cost, to study if these cost is helpful for improving communication quality is the mean direction of our study. In the study, functions that describe relationships between system performance and network size are monitoring in terms of average SINR, average capacity and area capacity under different reuse factor, base station transmission power and shadow fading variance value by Matlab simulation. Also in the worst condition is studied. We find in the research that higher density network can provide higher area capacity with the price of average SINR decreases while the system must be designed to fulfill the requirement of users in worst condition. To compromise these factors, the network density should be proper to both provide enough average SINR and area capacity.

The exponentially increasing demand for mobile broadband communications has led to the dense deployment of cellular networks with aggressive frequency reuse patterns. The future Fifth Generation (5G) networks are expected to overcome capacity and throughput challenges by adopting a multi-tier architecture where several low-power Base Stations (BSs) are deployed within the coverage area of the macro cell. Hence, Inter-Cell Interference (ICI) caused by the simultaneous usage of the same spectrum in different cells creates severe problems. ICI reduces system throughput and network capacity, and has a negative impact on cell-edge users and overall system performance. Therefore, effective interference coordination techniques are required, especially, for user-to-cell association and resource allocation to mitigate severe impact of ICI on system performance in 5G heterogeneous networks (HetNets). This is to improve Quality of Service (QoS) and maximize system throughput arising from the deployment of small cell overlay on macro BSs in heterogeneous cellular networks, which creates traffic load imbalance due to varying transmit power of different BSs in the downlink. In this research, a cell association scheme based on Cell Range Expansion (CRE), integrated with power control techniques is proposed. Simulation results are presented to show the ability of this technique to protect offloaded users from severe ICI and maximize throughput while achieving desirable QoS and load balancing for users of different tiers. With the advancement of information and computer technology, the envisioned 5G wireless communication is expected to encompass an unprecedented heterogeneous and ultra-dense communication environment. Vehicular communications play a vital role in 5G wireless network and have been widely studied recently due to its great potential to ensure reliability and support intelligent transportation and various safety applications. This research therefore exploits the tractability of stochastic geometry to analyze the coverage of urban vehicular networks, by deriving a closed-form expression to maximize the ergodic capacity of cellular users (CUEs) and mitigate interference, taking into consideration the QoS requirements of both vehicle-to-vehicle (V2V) and vehicleto-infrastructure (V2I) links. Consequently, the latency and reliability requirements of V2V/V2I links are formulated as optimization constraints, involving joint power allocation and spectrum sharing (PASS), taking into account the slow varying and large scale channel state information (CSI) measurements. Due to non-convex nature of the problem, the optimization is transformed into sub-optimal convex equivalence, while a low complexity Algorithm that yields optimal resource allocation is then designed to solve it. Simulation results are used to show enhanced performance in our approach compared to related works. Finally, the upsurge in the number of connected devices, such as smart cars, to the envisioned 5G technology is expected to pose high capacity and data rate demands on the network. The conventional access techniques (i.e., CDMA, TDMA and OFDMA) may not meet stringent requirements, such as ultra-low latency, high reliability, improved spectral efficiency and massive device connectivity. This work further investigates non-orthogonal multiple access (NOMA) technique as promising solution to improve spectral efficiency and reduce interference in 5G Ultra Dense Network (UDN). The NOMA scheme is combined with two promising capacity and bandwidth enhancement techniques - massive multiple input and multiple output (MIMO) and carrier aggregation (CA), for overall network performance. In particular, for the proposed novel NOMA-CA approach, we justify the importance of maintaining green communication as a key requirement for 5G with Energy Efficiency (EE) analysis. Firstly, a proportional fairness scheduler is used to perform resource allocation and maintain fairness among users based on their channel condition. Secondly, an optimization problem to maximize the EE weighted-sum under joint power and bandwidth allocation on each aggregated component carrier (CC) is formulated. Conventionally, the formulated optimization is transformed from non-convex to convex problem. An iteratively adaptive Algorithm is then developed to find optimal solution for the problem. Simulation results show better improvement in EE and sum rate compared to the traditional OMA scheme.

The envisioned rapid and exponential increase of wireless data traffic demand in the next years imposes rethinking current cellular networks due to the available spectrum scarcity. In this regard, 3 main drivers are considered to increase the capacity of today's most advanced (4G) and future (5G and beyond) cellular networks: use more bandwidth (more Hz) through spectral aggregation, enhance the spectral efficiency per base station (BS) (more bits/s/Hz/BS) by using multi-antenna (i.e. MIMO) systems, and increase the density of BSs (more BSs/km2) through a dense and heterogeneous deployment. We focus on the last 2 drivers. First, the use of MIMO systems allows exploiting the spatial dimension for improving the capacity of a conventional point-to-point link, increasing the number of served users, and reducing unwanted emissions (interference). Second, dense heterogeneous networks are a simple and cost-effective way to boost the area spectral efficiency by densifying the network and improving the spatial re-use of the spectrum. However, increasing the BSs density entails two main technical challenges: the interference increases because neighboring BSs/users are nearer and the amount of data traffic, as well as downlink (DL) and uplink (UL) traffic asymmetry, varies over space and time more drastically since the number of users per BS is reduced. The increase of interference makes the development of efficient interference management techniques a key enabler for MIMO dense heterogeneous networks. On the other hand, the variability of the per-BS traffic amount and the DL/UL traffic asymmetry convert flexible duplexing (i.e. flexible allocation of DL/UL resources per BS) into a necessity for an efficient resource usage. Therefore, the development of resource management schemes capable of adapting to the varying traffic load, as well as interference management, becomes crucial. Accordingly, this thesis focuses on the development of advanced interference management techniques to deal with inter-cell interference in MIMO dense networks and on the design of traffic- and interference-aware resource management schemes for flexible duplexing systems in asymmetric traffic conditions. To these goals, the wide deployment of MIMO systems is capitalized to develop advanced multi-antenna signal processing techniques when full reuse of time and frequency resources among densely deployed BSs is adopted. In the first part, different statistical characterizations of the transmitted signals are analyzed to improve the capacity of wireless interference-limited MIMO channels. Advanced signaling schemes are developed and the use of improper Gaussian signaling (IGS) is investigated, which allows exploiting the real and imaginary dimensions of MIMO channels. Majorization theory is exploited to demonstrate the strict superiority of IGS. In the second part, transmit coordination strategies are proposed to manage interference in extremely dense cellular networks. The design of BSs transmit strategies (involving design of spatial transmit/receive filters, power control, and user scheduling) is coordinated to optimize different network functions while reducing the stringent requirements needed for channel estimation in dense networks. Coordination strategies for the case in which different signaling schemes coexist in the network are also derived. Further, coordination strategies for cluster-based joint transmissions are developed, where BSs are grouped into clusters and different clusters interfere to each other. The third part focuses on the design of traffic- and interference-aware duplexing techniques to make a better use of the available resources by taking into account the asymmetric traffic conditions that arise in dense networks and managing the new kinds of interference that come up under flexible duplexing. Short-term and long-term optimizations are investigated, being therefore the interference managed instantaneously and statistically, respectively.
L'augment ràpid i exponencial previst per a la demanda de tràfic de dades en els pròxims anys imposa redissenyar les xarxes cel·lulars actuals degut a l'escassetat de l'espectre radioelèctric disponible. Es consideren 3 eixos directors per augmentar la capacitat dels sistemes més avançats d'avui dia (4G) i del futur (5G i més enllà): utilitzar més ample de banda (més Hz), millorar l'eficiència espectral per estació base (BS) (més bits/s/Hz/BS) utilitzant sistemes multi-antena (MIMO) i incrementar la densitat de BSs (més BSs/km2) a través d'un desplegament dens i heterogeni. Ens centrem en els 2 últims eixos. En primer lloc, l'ús de sistemes MIMO permet explotar la dimensió espacial per millorar la capacitat d'un enllaç convencional punt a punt, incrementar el nombre d'usuaris servits i reduir emissions indesitjades (interferències). En segon lloc, les xarxes denses i heterogènies són una manera simple i rentable de millorar l'eficiència espectral per àrea a través de la densificació de la xarxa i la reutilització espacial de l'espectre. No obstant això, l'increment de la densitat de BSs planteja dos principals reptes tècnics: les interferències augmenten perquè BSs/usuaris veïns estan més propers i la quantitat de tràfic de dades, així com l'asimetria del tràfic de baixada (DL) i de pujada (UL), fluctua amb el temps i l'espai més dràsticament ja que el nombre d'usuaris per BS és reduït. Per tant, un factor clau per a les xarxes MIMO denses i heterogènies és el desenvolupament de tècniques eficients de gestió d'interferències. D'altra banda, la variabilitat de la quantitat i asimetria del tràfic converteix en una necessitat el duplexat flexible (és a dir, assignacions flexibles de recursos DL/UL per BS) per aconseguir un ús eficient dels recursos. Així doncs, es torna crucial el desenvolupament d'esquemes de gestió de recursos capaços d'adaptar-se a càrregues de tràfic variable i, a la vegada, gestionar interferències. Aquesta tesi es centra en el desenvolupament de tècniques avançades de gestió d'interferències per combatre interferències entre cel·les en xarxes MIMO denses i en el disseny d'esquemes de gestió de recursos que tenen en compte el tràfic i la interferència per a sistemes de duplexat flexible en condicions asimètriques de tràfic. Per aconseguir aquests objectius, s'aprofita l'ampli desplegament de sistemes MIMO per desenvolupar tècniques avançades de processament de senyals quan s'adopta reutilització completa de recursos entre BSs densament desplegades. En la primera part, s'analitzen diferents caracteritzacions estadístiques dels senyals transmesos per millorar la capacitat dels canals limitats per interferència. Es deriven esquemes de senyalització avançats i s'investiga l'ús de la senyalització Gaussiana improper, la qual permet explotar les dimensions reals i imaginàries dels canals MIMO. En la segona part, es proposen estratègies de transmissió coordinades per gestionar interferències en xarxes denses. El disseny de les estratègies de transmissió a les BSs (incloent: disseny de filtres espacials en transmissió/recepció, control de potència i selecció d'usuaris) és coordinat per optimitzar diferents funcions de xarxa mentre que es redueixen els estrictes requisits d'estimació de canal en xarxes denses. També s'analitzen estratègies de coordinació per al cas en què diferents esquemes de senyalització coexisteixen. A més, es deriven estratègies de coordinació per a transmissions conjuntes basades en grups, on les BSs s'agrupen en grups i grups veïns s'interfereixen entre si. La tercera part es centra en el disseny de tècniques de duplexat flexible que tenen en compte tràfic i interferència per fer un millor ús dels recursos disponibles, considerant condicions de tràfic asimètriques i gestionant els nous tipus d'interferències que apareixen sota el duplexat flexible. S'investiguen optimitzacions a curt i a llarg termini, sent llavors la interferència gestionada instantàniament i estadísticament, respectivament.
El aumento rápido y exponencial previsto para la demanda de tráfico de datos en los próximos años impone rediseñar las redes celulares inalámbricas actuales debido a la escasez del espectro radioeléctrico disponible. En este sentido, se consideran tres ejes directores para aumentar la capacidad de las redes celulares más avanzadas de hoy en día (sistemas 4G) y las del futuro (sistemas 5G y más allá): - utilizar más ancho de banda (más Hz) a través de la agregación de espectro, - mejorar la eficiencia espectral por estación base (BS) (más bits/s/Hz/BS) utilizando múltiples antenas en las BSs y los usuarios (sistemas MIMO), e - incrementar la densidad de BSs (más BSs/km2) mediante un despliegue denso y heterogéneo (conocido como redes densas y heterogéneas). Esta tesis se centra en los dos últimos ejes directores. En primer lugar, el uso de sistemas multi-antena permite explotar la dimensión espacial con varias finalidades: mejorar la capacidad de un enlace inalámbrico convencional punto a punto, incrementar el número de usuarios servidos y reducir emisiones indeseadas (interferencias). En segundo lugar, las redes densas y heterogéneas son una manera simple y rentable de mejorar la eficiencia espectral por área a través de la densificación de la red con BSs de diferentes características y de la reutilización espacial del espectro radioeléctrico. Sin embargo, el incremento de la densidad de BSs plantea dos principales desafíos técnicos: - las interferencias en la red aumentan porque BSs/usuarios vecinos están más próximos y - la cantidad de tráfico de datos, así como la asimetría del tráfico de bajada (DL) y de subida (UL), fluctúa con el tiempo y el espacio más drásticamente debido a que el número de usuarios por BS se reduce. El aumento de interferencias en la red hace que un factor clave para las redes MIMO densas y heterogéneas sea el desarrollo de técnicas eficientes de gestión de interferencias. Pero, a medida que avanzamos hacia redes más densas, la gestión de interferencias se convierte cada vez en un reto más desafiante. Por otro lado, la variabilidad de la cantidad de tráfico de datos por BS y de la asimetría del tráfico DL/UL convierten en una necesidad el duplexado flexible (es decir, asignaciones flexibles y dinámicas de recursos DL/UL por BS, ya sea en el dominio temporal o frecuencial) para conseguir un uso eficiente de los recursos radio que satisfaga las cargas de tráfico no uniformes en espacio y variantes en tiempo. Por lo tanto, se vuelve crucial el desarrollo de esquemas de gestión de recursos capaces de adaptarse a cargas de tráfico variable y de, a su vez, gestionar las interferencias. En este sentido, esta tesis doctoral se centra en: 1. el desarrollo de técnicas avanzadas de gestión de interferencias para hacer frente a las interferencias entre celdas en redes celulares MIMO densas, y 2. el diseño de esquemas de gestión de recursos que tengan en cuenta el tráfico y la interferencia para sistemas de duplexado flexible bajo condiciones de tráfico asimétricas. Para alcanzar estos objetivos, se aprovecha el amplio despliegue de sistemas MIMO con el fin de desarrollar técnicas multi-antena avanzadas de procesado de señales cuando se adopta un reúso completo de los recursos en tiempo y en frecuencia entre BSs densamente desplegadas en la red. En la primera parte de la tesis, se analizan diferentes caracterizaciones estadísticas de las señales de transmisión para mejorar la capacidad de los canales inalámbricos interferentes. En este sentido, se desarrollan esquemas de señalización avanzados y se investiga el uso de la señalización Gaussiana improper (IGS), la cual permite aprovechar las dimensiones reales e imaginarias de los canales de propagación MIMO mediante la división de una dimensión espacial en dos mitades. La teoría de la majorización se explota para demostrar la superioridad estricta de IGS. Después, los beneficios de IGS se aplican a diferentes escenarios MIMO limitados por interferencia. Otra forma de gestionar la interferencia con reúso completo de los recursos frecuenciales es mediante la coordinación y/o cooperación de BSs. La coordinación entre BSs permite ajustar de manera coordinada las estrategias de transmisión de diferentes BSs con el objetivo de reducir el impacto de las interferencias en la red. Por el contrario, la cooperación entre BSs permite que las BSs actúen como un único transmisor multi-antena y tiene la gran ventaja de que convierte la interferencia en señal útil a través de la transmisión conjunta de BSs cooperativas hacia un mismo usuario. Sin embargo, la cooperación requiere sincronización estricta y alta capacidad de backhaul para compartir datos de usuario entre BSs. Por esta razón, en implementaciones prácticas, la cooperación sólo se puede lograr entre un número reducido de BSs (las cuales forman un grupo) y la coordinación entre grupos sigue siendo necesaria para hacer frente a las interferencias. Tanto la coordinación como la cooperación, ya sean implementadas de forma centralizada o descentralizada, requieren el conocimiento de todos los canales de propagación de la red, lo cual impone requisitos estrictos en cuanto a estimación de canal para la gestión de interferencias en redes densas. En la segunda parte de este trabajo se proponen estrategias de transmisión coordinadas para gestionar interferencias en las redes celulares extremadamente densas. El foco está en la transmisión DL. El diseño de las estrategias de transmisión en las BSs (incluyendo el diseño de los filtros espaciales de transmisión y recepción, el control de potencia y la selección de usuarios) es coordinado con tal de optimizar diferentes funciones de red (como, por ejemplo, la suma ponderada de las tasas de transmisión), mientras que se reducen los estrictos requisitos necesarios para estimación de canal en redes densas. También se analizan estrategias de coordinación para el caso en que diferentes esquemas de señalización (proper e improper) coexisten en la red. Además, la tesis deriva estrategias de coordinación para transmisiones conjuntas basadas en grupos, donde las BSs se agrupan en grupos formados por un número reducido de BSs cooperativas y grupos vecinos se interfieren entre sí. En este caso, la estrategia de transmisión se optimiza conjuntamente con la formación de los grupos. Por último, se aborda la gestión de recursos en sistemas de duplexado flexible, donde los recursos tienen que ser distribuidos adecuadamente entre las transmisiones DL y UL de acuerdo con las asimetrías y la cantidad de tráfico de cada BS. Bajo una reutilización de recursos en BSs densamente desplegadas, el uso del duplexado flexible conlleva cambios en la interferencia generada entre BSs y/o usuarios vecinos. Como consecuencia, surgen nuevos tipos de interferencias (como la interferencia de BS a BS). La tercera parte de la tesis se centra en el diseño de técnicas de duplexado flexible que tienen en cuenta el tráfico para la gestión de recursos y de interferencias. En contraste con las partes anteriores, se consideran transmisiones DL y UL para cada BS. El objetivo principal es hacer un mejor uso de los recursos tiempo/frecuencia disponible, teniendo en cuenta las condiciones de tráfico asimétricas que surgen en redes densas, así como la gestión de los nuevos tipos de interferencias que aparecen bajo sistemas de duplexado flexible. Se investigan optimizaciones a corto plazo y a largo plazo, siendo entonces la interferencia gestionada de manera instantánea y de manera estadística, respectivamente.

The current decades have witnessed the explosive increase of traffic-data demand. It is predicted that indoor wireless communications will be one of the fastest growing markets, since the vast majority (over 80%) of data demand occurs in indoors. Facing such a huge data demand, the dense deployment of small cells (SCs) in indoor environments is boosted, which brings breakthroughs of throughput for in-building communications. However, the densification of indoor small-cell (SC) networks also poses new challenges, such as complex propagating environments, severe blockage effects and short link distances, which significantly influence the evaluation of network performance. This thesis mainly investigates the performance analysis of indoor dense SC networks. Firstly, the probability of Line-of-Sight (LOS) propagation is crucial to model the real signal propagation channels and to evaluate the performance of cellular networks. However, existing LOS probability models are oversimplified to provide the exact LOS probability in indoor scenarios. By considering the realistic layout of building structures, this thesis proposes a novel and analytical LOS probability model for downlink radio propagations in typical indoor scenarios, which have rectangular rooms and corridors. Through the proposed model, the LOS probability can be calculated directly without the measurement and simulation. Next, in terms of the impact of LOS and Non-Line-of-Sight (NLOS) transmissions, the traditional works do not distinguish them, which is not practical for dense cellular networks. Thus, a tractable path loss model considering both LOS and NLOS propagations is proposed for the performance analysis of indoor dense SC networks. Based on the theory of stochastic geometry, the performance metrics, such as coverage probability, spectral efficiency (SE) and area spectral efficiency (ASE), are analytically derived. The analytical results provide insights into the design of indoor dense SC networks in the future. Thirdly, regarding the severe effects of blockages in indoor environments, the traditional approach that simply considers it as a log-normal shadowing is too simple. Therefore, a wall blockage model is developed to characterize the impact of blockages based on the stochastic geometry. Furthermore, the mathematical expression of coverage probability for the case of impenetrable blockages is derived, which employs a path loss model incorporating both the blockage-based and distance-based path loss.

Avec le développement rapide des nanoréseaux électromagnétiques denses, il existe un besoin urgent des protocoles de routage spécialisés capables de relever les défis uniques de ces environnements ultra-denses. Les dispositifs à l’échelle nanométrique offrent une bande passante élevée et sont confrontés à des complexités de communication importantes.Cette thèse aborde le manque de protocoles de routage viables pour les nanoréseaux denses en développant et en évaluant un nouvel algorithme spécifiquement conçu pour ce contexte. Une contribution clé est l’introduction du mécanisme EIDA (Equitable Distributed ID Assignment), qui équilibre les compromis entre les attributions d’id idéales et aléatoires dans ces réseaux. EIDA est ensuite intégré au protocole FR-SLR (Forwarder Reduction in SLR Routing), ou il sert de mécanisme d’attribution d’id. FR-SLR optimise le protocole SLR existant en réduisant le nombre de redirecteurs de paquets et en garantissant une transmission équitable des paquets au meilleur effort.De plus, cette thèse explore l’impact de divers paramètres réseau (β, portée de communication, densité et durée d’impulsion) sur la qualité de la communication (nombre de collisions de paquets, réception, émission et livraison), fournissant des informations précieuses pour sélectionner les paramètres réseau appropriés. Elle s’intéresse également aux mécanismes de concurrence tels que l’interférence constructive et l’effet de capture, en identifiant les niveaux de concurrence qui peuvent être efficacement pris en charge dans les nanoréseaux (au niveau d’impulsion, du bit, du paquet, et du flux). En raison des défis actuels dans la fabrication de nanomachines, le protocole, le mécanisme et l’évaluation proposés ont été e validés par des simulations a l’aide de BitSimulator, un simulateur permettant la simulation de nanoréseaux ultra-denses
Wireless networks are constantly evolving and are now ubiquitous in society. They meet new needs, which are very promising in terms of job creation. Their evolution is also illustrated by the recent creation of the master's degree "Internet of Things", cohability by the UBFC and UTBM, and carried by our research team. In particular, there has been an explosion in the number of terminals (users or autonomous equipment) and consequently, in the number of equipment with direct communication range, such a dense network can have direct neighborhoods of up to thousands or even tens of thousands of nodes. However, conventional protocols and algorithms are not adapted to this situation, as strategies have so far prevailed, organizing communications in relatively small, frozen cells (e.g. a relay controlling a number of slave terminals) and dedicated to specific tasks. The current thesis aims to design, implement and analyze new algorithms and protocols for dense networks, in particular, to allow much less fixed communication patterns (so-called "Many to many") and less costly (by at least partially starving themselves of binding base stations). In particular, there is a need to note a cross-layer approach and taking into account the specificities of these networks in all their protocol layers, although we will be particularly interested in routing and transport levels. These new methods open up new perspectives in many existing applications and even make new ones possible, such as programmable material. We can really talk about a paradigm shift for communication in networks

Dense wavelength division multiplexing (DWDM) is becoming the technology of choice for meeting the increasing bandwidth demands in optical networks. DWDM has been used to increase the capacity of long-haul optical transport systems. Efforts are being made to move DWDM into the broadband access network serving residential and business subscribers. First, a new centralized DWDM PON scheme is demonstrated for bi-directional upstream and downstream transmissions. The proposed DWDM PON scheme is implemented using optical carrier suppression and separation (OCSS) technology to generate a wavelength pair from a single laser source at the central office. This method enables the co-location of both upstream and downstream DWDM transmitters in the central office. In addition, the complexity, cost, and maintenance of the optical network unit are reduced by enabling wavelength independent operation. Second, a new multistage architecture is proposed for the delivery of information to groups of subscribers located at different distances from the central office. A 25 GHz DWDM comb is generated using OCSS technology, and error-free transmission of four 10 Gbps channels is demonstrated. Finally, a new wide area access network with bi-directional DWDM amplification using semiconductor optical amplifiers (SOAs) is demonstrated. The detrimental effect of SOA crosstalk resulting from cross gain modulation can be suppressed using a constant intensity modulation format such as differential phase shift keying (DPSK). The feasibiity of bi-directional DPSK transmission of 16 interleaved DWDM channels using an in-line SOA has been studied experimentally. In addition, the reduction of bi-directional SOA reflections has been realized by optimizing the SOA bias current and facet reflectivities.

Both data traffic and number of subscriptions have enormously increased inmobile network in recent years. Moreover, there will be an even faster growthin the future. A promising way to satisfy a significantly increasing demand infuture radio access network is by using so called Ultra Dense Networks (UDNs)which deploy a large number of base stations compared to the number of activeusers.Radio spectrum is a finite resource and therefore has to be shared by multipleusers. This sharing of radio spectrum inevitably causes interference betweenthe users. In this study, the interference management performance of differentresource allocation schemes in different network density is studied, which is froma traditional network density to ultra dense network.Except for traditional frequency reuse scheme and reuse partitioning scheme,Coordinated Multi Point (CoMP) schemes have been chosen in the work. Different CoMP techniques such as the universal frequency reuse (UFR) and coop-erative frequency reuse (CFR) are tested to find the best network performancein terms of average users data throughput and cell rate.Besides, after measuring these CoMP schemes which are designed for highbase station density, the optimal scheme is found to be a potential methodadopted by ultra-dense network.

This thesis examines the problem of downlink power allocation in dense 5Gnetworks, and attempts to develop a data-driven solution by employing deepreinforcement learning. We train and test multiple reinforcement learningagents using the deep Q-networks (DQN) algorithm, and the so-called Rainbowextensions of DQN. The performance of each agent is tested on 5G UrbanMacro simulation scenarios, and is benchmarked against a fixed power allocationapproach. Our test results show that the DQN models are successful atimproving data rates at cell-edge, while generalizing well to previously unseensimulation scenarios. In addition, the agents induce throughput balancing effects,i.e., achieve fairness among users, in networks with full-downlink-buffertraffic by properly designing the reward signal.
Det här examensarbetet undersöker kraftallokering i nedlänksriktning för täta5G-nätverk och försöker utveckla en datadriven lösning genom användning avdeep reinforcement learning. Vi tränar och testar flera reinforcement learningagentermed deep Q-networks (DQN) algoritmen, och de så kallade ”Rainbowextensions” av DQN. Prestandan av varje agent testas på storskaliga tätortsscenarionför 5G, och jämförs med en fast kraftallokeringsmetod.Våra testresultatvisar att DQN-modellerna leverar högre överföringshastigheter vid cellkanten,samtidigt som metoden fungerar väl för okända simuleringsscenarion. Utöverhastighetsökningen så balanserar agenterna dataflödet, vilket leder till rättvisallokering bland användarna i nätverk med ”full-downlink-buffer”-trafik genomatt korrekt designa belöningssignalen.

Les protocoles de communications à accès aléatoires sont des candidats prometteurs pour les futurs systèmes de communications sans fil dédiés aux applications machine à machine (M2M). Ces méthodes d’accès sont généralement basées sur des techniques d'accès aléatoires mettant en œuvre des concepts simples de sondage de canal et de report de la transmission pour réduire les collisions, tout en évitant l'utilisation d'ordonnanceurs complexes. Parmi les différents protocoles, Carrier sense multiple access/collision avoidance with a Request-To-Send/Clear-To-Send (CSMA/CA-RTS/CTS) est un protocole qui pourrait être adopté pour les scénarios de M2M. Cette approche est efficace pour éviter les collisions entre les paquets de données. Cependant dans le cas d’un réseau très dense, les performances sont dégradées à cause de la forte probabilité de collisions. Pour atténuer cet effet, les collisions entre les messages de contrôles RTS doivent être réduites. Cette thèse propose de résoudre ce problème en divisant le canal commun en sous-canaux pour transmettre les messages de contrôle de demande d’accès au canal ; le canal commun est utilisé dans son ensemble pour la transmission de données. L’ajout d’un degré de liberté pour le message de demande d’accès permet de réduire la probabilité de collision, et donc d’améliorer les performances du système notamment dans des scénarios avec des nombres importants de nœuds souhaitant communiquer. Dans ce travail, nous dérivons ainsi une solution complète de méthode d’accès en s'appuyant sur le CSMA / CA - RTS / CTS et en multiplexant une configuration multi-canal pour les messages RTS et un canal unique pour la transmission de données. Une version améliorée, basée sur l'ordonnancement des utilisateurs, est également étudiée. Un modèle analytique a été développé, analysé et validé par simulations. Celui-ci est une extension du modèle Bianchi. Les performances en termes de débit saturé, de temps de transmission et de la probabilité de rejet de paquets sont discutées. Enfin, les impacts liés à la prise en compte d’une couche physique de type multi porteuses sont discutés dans le dernier chapitre
Opportunistic protocols are promising candidates for future wireless systems dedicated to machine to machine (M2M) communication. Such protocols are usually based on a random access with simple techniques of medium sensing and deferring to reduce collisions while avoiding the use of complex schedulers. Among different protocols, Carrier sense multiple access/collision avoidance with a Request-To-Send/Clear-To-Send (CSMA/CA-RTS/CTS) is an opportunistic protocol which could be adopted for M2M scenarios. Such approach is efficient to avoid collisions between data packets but in a very dense network, the random access used to send the RTS suffers itself from a high probability of collision which degrades the performance. In order to mitigate this effect, RTS collisions should be reduced. This thesis proposes to address this issue by splitting the common channel in sub-channels for transmitting the RTS messages. While the common channel is used as a whole for data transmission. Multiple nodes can then contend in time and frequency for these RTS sub-channels, thereby reducing RTS collisions and increasing overall efficiency. In this work, we thus derive a complete protocol solution relying on CSMA/CA - RTS/CTS multiplexing a multi-channel configuration for RTS messages and a unique channel for data transmission. An enhanced version based on users scheduling is integrated as well. In this thesis, the proposed protocol is investigated from a joint PHY-MAC point of view. This strategy is shown to provide better system performance particularly for loaded networks. An accurate analytical model derived as a straightforward extension of the Bianchi model is analyzed and validated by simulations. Performance in terms of saturation throughput, transmission delay and packet drop probability is discussed

The traffic volume in wireless communication has grown dramatically in the last decade and is predicted to keep increasing in the future. In this thesis, we focus on the densification dimension for capacity improvement, which has been proved to be the most effective in the past. The current gain of network densification mainly comes from cell splitting, thereby serving more user equipments (UEs) simultaneously. This trend will decelerate as base station (BS) density gets closer to or even surpass UE density which forms an ultra-dense network (UDN). Thus, it is crucial to understand the behavior of ultra-densification for future network provisioning.   We start from comparing the effectiveness of densification with spectrum expansion and multi-antenna systems. Our findings show that deploying more BSs provides a substantial gain in sparse network but the gain decreases progressively in a UDN. The diminishing gain appears in a UDN make us curious to know if there exists a terminal on the way of densification. Such uncertainty leads to the study on the asymptotic behavior of densification. We incorporate a sophisticated bounded dual-slope path loss model and practical UE densities in our analysis. By using stochastic geometry, we derive the expressions and prove the convergence of the coverage probability of a typical UE and network area spectral efficiency (ASE). Considering the large portion of dormant BSs in a UDN, it is an interesting question whether we can utilize these dormant BSs to improve system performance is an interesting question. To this end, we employ joint transmission (JT) techniques into a UDN. Two types of cooperation schemes are investigated: non-coherent JT and coherent JT depending on the availability of channel state information (CSI). Our results reveal that non-coherent JT is not beneficial in a UDN while coherent JT are able to increase UE spectral efficiency (SE) depending on the environmental parameters.

QC 20170117

Malgré leur réussite remarquable, les premières versions des normes de réseaux locaux sans fil IEEE 802.11, IEEE 802. 11 a/b/g WLAN, sont caractérisées par une efficacité spectrale faible qui est devenue insuffisante pour satisfaire la croissance explosive de la demande de capacité et de couverture. Grâce aux progrès considérables dans le domaine des communications sans fil et l'utilisation de la bande de fréquence autour de 5 gigahertz le standard IEEE 802.11n et plus récemment 1'IEEE 802.11ac ont amélioré les débits offerts par la couche physique. Cela été possible grâce principalement à l'introduction des techniques multi-antennaires (MIMO, pour Multiple-Input) et des techniques avancées de modulation et de codage. Aujourd'hui, deux décennies après sa première apparition, le Wi-Fi est présenté comme une technologie WLAN permettant des débits supérieurs à 1 gigabit par seconde. Cependant, dans la plupart des scénarios de déploiement du monde réel, il n'est pas possible d'atteindre la pleine capacité offerte par la couche physique. Avec la croissance rapide de la densité des déploiements des WLANs et l'énorme popularité des équipements Wi-Fi, la réutilisation spatiale doit être optimisée. D'autre part, des nouveaux cas d’utilisation sont prévus pour décharger les réseaux cellulaires et pour couvrir des grandes surfaces (stades, gares, etc.). Ces environnements de haute densité représentent un vrai défi pour les générations actuelles de Wi-Fi qui doivent offrir une meilleure qualité à moindre coût. C'est dans ce contexte que s’inscrit l'objectif de cette thèse qui porte sur l'amélioration de l'efficacité des protocoles de la couche MAC des réseaux WLAN de haute densité. Notamment, un des buts de cette thèse est de contribuer à la préparation de la prochaine génération du standard Wi-Fi : IEEE 802.11ax High Efficiency WLAN (HEW). Plutôt que de continuer à cibler l'augmentation des débits maximums théoriques, nous nous concentrons dans le contexte de HEW sur l'amélioration du débit réel des utilisateurs. Pour cela, on prend en compte tous les autres équipements associés à des WLANs voisins, qui essayent d'accéder au même canal de transmission d’une manière simultanée. Pour améliorer la performance du Wi-Fi dans ces environnements denses, nous proposons une adaptation dynamique du mécanisme de détection de signal. Comparé au contrôle de la puissance de transmission, le mécanisme proposé est plus incitatif parce que l'utilisateur concerné bénéficie directement de son application. Les résultats de nos simulations montrent des gains importants en termes de débit atteint dans les scénarios de haute densité. Ensuite, nous étudions l’impact de la nouvelle adaptation sur les mécanismes de sélection de débit actuellement utilisés. D'après les résultats obtenus, 1'adaptation proposée peut être appliquée sans avoir besoin de modifications substantielles des algorithmes de sélection de débit. Pour améliorer l'équité entre les différents utilisateurs, nous élaborons une nouvelle approche distribuée pour adapter conjointement le mécanisme de détection de signal et le contrôle de la puissance de transmission. Cette approche est évaluée ensuite dans différents scénarios de simulation de haute densité où elle prouve sa capacité à résoudre les problèmes d'équité en particulier en présence de nœuds d'anciennes générations dans le réseau, cela tout en améliorant le débit moyen d'un facteur 4 par rapport à la performance conventionnelle du standard. Enfin, nous concevons et mettons en œuvre une solution centralisée basée sur l'apprentissage à base de réseaux de neurones. Cette approche repose sur l'adaptation conjointe de puissance de transmission et du mécanisme de détection du signal. [...]
Despite their remarkable success, the first widely spread versions of the Institute of Electrical and Electronics Engineers (IEEE) 802.11 Wireless Local Area Network (WLAN) standard, IEEE 802. 11 a/b/g, featured low spectral efficiencies that are becoming insufficient to satisfy the explosive growth in capacity and coverage demands. Thanks to the advances in the communication theory and the use of the 5 GHz frequency band, the IEEE 802.11n and recently the IEEE 802.1lac amendments improved the Physical Layer (PHY) data rates by introducing Multiple-Input Multiple Output (MIMO) techniques, higher Modulation and Coding Scheme (MCS), etc. Today, after almost two decades of its first appearance, Wi-Fi is presented as a gigabit wireless technology. However, the full potential of the latest PHY layer advances cannot be enabled in all real world deployment scenarios. With the rapidly increasing density of WLAN deployments and the huge popularity of Wi-Fi enabled devices, spatial reuse must be optimized. On another hand, the new challenging use case environments and the integration of mobile networks mainly for cellular offloading are limiting the opportunity of the current Wi-Fi generations to provide better quality at lower cost.In this thesis, we contribute to the current standardization efforts aiming to leverage the Wi-Fi efficiency in high density environments. At the time of writing this document, the IEEE 802.11ax Task Group (TG) is developing the specification for the High Efficiency WLAN (HEW) standard (next Wi-Fi evolution). Rather than continuing to target increased theoretical peak throughputs, we focus in the context of HEW on improving the throughput experienced by users in real life conditions where many other devices, belonging to neighboring overlapping networks, simultaneously contend to gain access. To enhance this performance, we propose a dynamic adaptation of the carrier sensing mechanism. Compare to controlling the transmission power, the proposed mechanism has more incentives because it benefits directly the concerned user. Extensive simulation results show impor1ant throughput gains in dense scenarios. Then, we study the impact of the new adaptation on the current rate control algorithms. We find that our adaptation mechanism operates efficiently without substantially modifying these algorithms that are widely used in today's operating WLANs. Furthermore, after analyzing the fairness performance of the proposed adaptation, we devise a new approach to jointly adapt the carrier sensing and the transmission power in order to preserve higher fairness degrees while improving the spatial reuse. This approach is evaluated in different dense deployment scenarios where it proves its capability to resolve the unfairness issues especially in the presence of legacy nodes in the network, while improving the achieved throughput by 4 times compared to the standard performance. Finally, we design and implement centralized learning-based solution that uses also an approach based on joint adaptation of transmission power and carrier sensing. This new solution takes benefit from the capability of artificial neural networks to model complex nonlinear functions to optimize the spatial reuse in dense WLANs while preserving fairness among contending nodes. The different contributions of this work have helped bring efficient solutions for future WiFi networks. We have presented these solutions to the IEEE 802.11ax TG where they were identified as important potential technical improvements for the next WLAN standard

The appetite for wireless high-data rate services is expected to continue for many years to come and drive the need for more capacity. Ultra-dense networks (UDNs) represent a paradigm shift where each base station (BS) serves only a few user equipments (UEs). By most accounts, most of the traffic will be generated indoor and operate in time-division duplex (TDD). This thesis considers dynamic TDD which has shown to perform well indoor for fluctuating traffic where the shorter communication range enables similar transmit powers to be used in uplink and downlink, but also generates potentially more harmful same-entity interference. Because of the sheer number of cells in UDN, the interference management needs to be both effective and scalable.   In the first part of the thesis, we compare static TDD with non-cooperative dynamic TDD and show that flexible time resource allocation is preferred for indoor UDNs. However, since it only provides a lower bound on performance, additional interference coordination is required. Unfortunately, existing schemes often consider either too few, too many, or simply the wrong interferers. We introduce a scheduling model that relates BS-to-BS interferences measured offline to individual BS activation probability taking into account traffic and propagation environment. Results show that the proposed scheme performs well when interference is high, and optimally when interference is low.   In the second part, we introduce cooperation to utilize the otherwise idle BSs and mitigate same- and other entity interference. Zero forcing (ZF) is employed in the downlink where not only downlink UEs but also uplink BSs are included in the precoding. Since downlink BSs do not know the information to be sent by uplink UEs beforehand, dummy symbols with zero power are transmitted. It shown that both uplink and downlink performance improves at low and medium load. Furthermore, it is possible to trade performance in the two directions at high load.

QC 20170922

Le réseau sans fil est l'un des domaines de réseautage les plus prometteurs avec des caractéristiques uniques qui peuvent fournir la connectivité dans les situations où il est difficile d'utiliser un réseau filaire, ou lorsque la mobilité des nœuds est nécessaire. Cependant, le milieu de travail impose généralement diverses contraintes, où les appareils sans fil font face à différents défis lors du partage des moyens de communication. De plus, le problème s'aggrave avec l'augmentation du nombre de nœuds. Différentes solutions ont été introduites pour faire face aux réseaux très denses. D'autre part, un nœud avec une densité très faible peut créer un problème de connectivité et peut conduire à l'optension de nœuds isolés et non connectes au réseau. La densité d'un réseau est définit en fonction du nombre de nœuds voisins directs au sein de la portée de transmission du nœud. Cependant, nous croyons que ces métriques ne sont pas suffisants et nous proposons une nouvelle mesure qui considère le nombre de voisins directs et la performance du réseau. Ainsi, la réponse du réseau, respectant l'augmentation du nombre de nœuds, est considérée lors du choix du niveau de la densité. Nous avons défini deux termes: l'auto-organisation et l'auto-configuration, qui sont généralement utilisés de façon interchangeable dans la littérature en mettant en relief la différence entre eux. Nous estimons qu'une définition claire de la terminologie peut éliminer beaucoup d'ambiguïté et aider à présenter les concepts de recherche plus clairement. Certaines applications, telles que Ies systèmes "In-Flight Entertainment (IFE)" qui se trouvent à l'intérieur des cabines d'avions, peuveut être considérées comme des systèmes sans fil de haute densité, même si peu de nœuds sont relativement présents. Pour résoudre ce problème, nous proposons une architecture hétérogène de différentes technologies à fin de surmonter les contraintes spécifiques de l'intérieur de la cabine. Chaque technologie vise à résoudre une partie du problème. Nous avons réalisé diverses expérimentations et simulations pour montrer la faisabilité de l'architecture proposée. Nous avons introduit un nouveau protocole d'auto-organisation qui utilise des antennes intelligentes pour aider certains composants du système IFE; à savoir les unités d'affichage et leurs systèmes de commande, à s'identifier les uns les autres sans aucune configuration préliminaire. Le protocole a été conçu et vérifié en utilisant le langage UML, puis, un module de NS2 a été créé pour tester les différents scénarios
Wireless networking is one of the most challenging networking domains with unique features that can provide connectivity in situations where it is difficult to use wired networking, or when ! node mobility is required. However, the working environment us! ually im poses various constrains, where wireless devices face various challenges when sharing the communication media. Furthermore, the problem becomes worse when the number of nodes increase. Different solutions were introduced to cope with highly dense networks. On the other hand, a very low density can create a poor connectivity problem and may lead to have isolated nodes with no connection to the network. It is common to define network density according to the number of direct neighboring nodes within the node transmission range. However, we believe that such metric is not enough. Thus, we propose a new metric that encompasses the number of direct neighbors and the network performance. In this way, the network response, due to the increasing number of nodes, is considered when deciding the density level. Moreover, we defined two terms, self-organization and self-configuration, which are usually used interchangeably in the literature through highlighting the difference ! between them. We believe that having a clear definition for terminology can eliminate a lot of ambiguity and help to present the research concepts more clearly. Some applications, such as In-Flight Entertainment (IFE) systems inside the aircraft cabin, can be considered as wirelessly high dense even if relatively few nodes are present. To solve this problem, we propose a heterogeneous architecture of different technologies to overcome the inherited constrains inside the cabin. Each technology aims at solving a part of the problem. We held various experimentation and simulations to show the feasibility of the proposed architecture

In recent years, the increase in the population of mobile users and the advances in computational capabilities of mobile devices have led to an exponentially increasing traffic load on the wireless networks. This trend is foreseen to continue in the future due to the emerging applications such as cellular Internet of things (IoT) and machine type communications (MTC). Since the spectrum resources are limited, the only promising way to keep pace with the future demand is through aggressive spatial reuse of the available spectrum which can be realized in the networks through dense deployment of small cells. There are many challenges associated with such densely deployed heterogeneous networks (HetNets). The main challenges which are considered in this research work are capacity enhancement, velocity estimation of mobile users, and energy efficiency enhancement. We consider different approaches for capacity enhancement of the network. In the first approach, using stochastic geometry we theoretically analyze time domain inter-cell interference coordination techniques in a two-tier HetNet and optimize the parameters to maximize the capacity of the network. In the second approach, we consider optimization of the locations of aerial bases stations carried by the unmanned aerial vehicles (UAVs) to enhance the capacity of the network for public safety and emergency communications, in case of damaged network infrastructure. In the third approach, we introduce a subsidization scheme for the service providers through which the network capacity can be improved by using regulatory power of the government. Finally, we consider the approach of device-to-device communications and multi-hop transmissions for enhancing the capacity of a network. Velocity estimation of high speed mobile users is important for effective mobility management in densely deployed small cell networks. In this research, we introduce two novel methods for the velocity estimation of mobile users: handover-count based velocity estimation, and sojourn time based velocity estimation. Using the tools from stochastic geometry and estimation theory, we theoretically analyze the accuracy of the two velocity estimation methods through Cramer-Rao lower bounds (CRLBs). With the dense deployment of small cells, energy efficiency becomes crucial for the sustained operation of wireless networks. In this research, we jointly study the energy efficiency and the spectral efficiency in a two-tier HetNet. We optimize the parameters of inter-cell interference coordination technique and study the trade-offs between the energy efficiency and spectral efficiency of the HetNet.

With rapidly increasing traffic demand, it is expected that ultra-dense wireless access networks are deployed in many buildings in a near future. Performance evaluation of in-building ultra-dens networks is thus of profound importance. Buildings consist of walls and floors in three-dimensional environments, and the walls and floors attenuate the radio propagation. However, previous studies on the performance evaluation of wireless networks have mainly focused on open areas with an assumption of two-dimensional environments.  In this thesis, we investigate the effects of walls and floors on the performance of user data rate when wireless access networks are densely deployed inside a building. We assume a building of a typical shape, and perform Monte Carlo simulations with multiple configurations of different wall and floor losses as well as different sets of numbers of users and base stations per floor. Numerical results indicate that penetration loss due to walls and floors can increase the data rate of both average and five-percentile users, as this tends to better isolate a given base station and its connected users from the signals of others. We also observe that increasing the number of indoor base stations does not necessarily improve received user data rate because the number of users is limited

Recentemente è stata notata una crescita dei servizi multimediali richiesti dagli utenti finali; in tal modo numerose soluzioni sono state implementate per garantire elevati bit rate e qualità del servizio necessari per questo tipo di applicazioni. Le reti completamente ottiche sono state stese in molte nazioni (Giappone, Corea, Cina) per fornire servizi a banda larga fino a casa dell'utente. Conseguentemente, sono richiesti dispositivi in grado di operare nel dominio ottico in modo tale da evitare il noto “collo di bottiglia” derivante dalle conversioni di formato O/E/O (ottico/elettrico/ottico). In questo modo, nuovi tipi di sistemi (per esempio: optical processing e passive optical network) in grado di operare nel dominio completamente ottico sono richiesti poiché solo questo tipo di soluzione è la miglior strada per offrire alte prestazioni in termini di servizi, rate e riduzione dei costi per bit. Il lavoro eseguito durante questo dottorato di ricerca è stato incentrato sull'evoluzione di dispositivi per la rigenerazione ottica in grado di operare al contempo una Ri-amplificazione e Ri-sagomatura (2R) dei segnali ottici. Studi ed esperimenti sono stati effettuati nei laboratori dell’ ISCOM sfruttando la possibilità di rigenerazione completamente ottica di un dispositivo 2R multi-canale in grado di lavorare e gestire più clients nel medesimo istante temporale. Il sistema è stato implementato in uno scenario DWDM (Dense Wavelegth Division Multiplexing); inoltre, lavorando nel dominio completamente ottico sono state eliminate le conversioni di formato (O/E/O). Il sistema di rigenerazione è basato sulla modulazione di fase presente all'interno della fibra ottica usata per ottenere, sotto particolari condizioni, la generazione di nuove repliche del segnale originario che si vuole rigenerare. Queste nuove repliche, essendo posizionate a nuove lunghezze d’onda, possono essere usate sia per ottenere una conversione di lunghezza d’onda sia per ottenere una rigenerazione ottica dei segnali. Ciascuna replica, infatti, è caratterizzata dall’ avere un andamento simile alle funzioni di Bessel in grado di eliminare il rumore accumulatosi durante la trasmissione dei segnali. L’idea di questo lavoro è basato su un approccio multi-lunghezza d’onda in modo tale da poter usare un solo dispositivo per fornire una rigenerazione 2R completamente ottica ai numerosi utenti operanti a 10 Gbps. La capacità dei sistemi, implementati nei laboratori ISCOM, di risagomare i segnali, è stata confermata sperimentalmente in termini di misurazioni di diagramma ad occhio dei segnali di uscita e dalle curve di BER (Bit Error Rate).
A recent increase of multimedia service demand from end-users has been noticed, thus several solutions have been implemented to guarantee the high rate and relative QoS (Quality of Service) needed for these kind of services. All optical networks have been deployed in many countries (Japan, Korea, China, at all) in order to supply broadband services to the home. Consequently, devices able to operate in optical domain are requested in order to avoid the so called “bottle-neck” coming from the O/E/O data conversion format. Thus, new kind of systems (optical processing and passive optical networks, at all) able to operate in photonic domain are requested because only this kind of solution is the better way to offer high performances in term of services, rate and low cost per bit. The work performed during this PhD program has been focused on the evolution of regeneration devices able to perform Re-amp and Re-shaping also know as 2R. Studies and experiments have been carried out at the ISCOM labs exploiting the possibility to a multi-channel 2R all optical regeneration device which is able to work with different client signals at the same time. The system has been implemented in a dense WDM (Wavelength Division Multiplexing) scenario. Moreover, working completely in optical domain, the format conversion (O/E/O) is avoided. The regeneration system is based on phase modulation present in the fiber and used to obtain, under particular conditions, the generation of new signal replica. These new replica, being placed at new different wavelengths can be used both to reach a wavelength conversion and to obtain an all optical regeneration effect. Each replica, in fact, is characterized by a Bessel like transfer function able to clean the noise accumulated along the signal transmission. The idea of this work is based on a multi-wavelength approach, thus only one device can be used to provide all optical 2R regeneration to several client signals at 10 Gbps at the same time. The ability of the systems, implemented at the ISCOM labs, to reshape the signals, has been experimentally confirmed in terms of eyes diagrams and BER (Bit Error Rate) measurements.

The growing number of wireless devices and wireless systems present many challenges on the design and operation of these networks. We focus on massively dense ad hoc networks and cellular systems. We use the continuum modeling approach, useful for the initial phase of deployment and to analyze broad-scale regional studies of the network. We study the routing problem in massively dense ad hoc networks, and similar to the work of Nash, and Wardrop, we define two principles of network optimization: user- and system-optimization. We show that the optimality conditions of an appropriately constructed optimization problem coincides with the user-optimization principle. For different cost functions, we solve the routing problem for directional and omnidirectional antennas. We also find a characterization of the minimum cost paths by extensive use of Green's theorem in directional antennas. In many cases, the solution is characterized by a partial differential equation. We propose its numerical analysis by finite elements method which gives bounds in the variation of the solution with respect to the data. When we allow mobility of the origin and destination nodes, we find the optimal quantity of active relay nodes. In Network MIMO systems and MIMO broadcast channels, we show that, even when the channel offers an infinite number of degrees of freedom, the capacity is limited by the ratio between the size of the antenna array at the base station and the mobile terminals position and the wavelength of the signal. We also find the optimal mobile association for the user- and system-optimization problem under different policies and distributions of the users.

: The increased interest in traffic-intensive applications such as High Definition (HD) video, augmented reality, and 3-D visualization is expected to result in higher network traffic. Such higher-fold traffic growth requires a significant paradigm shift in implementing upcoming 5G technology so that the user requests can be accommodated at the core network without causing a bottleneck. Emerging mobile content caching techniques can efficiently relieve overloaded network by caching popular content at intermediate nodes and user devices. Its efficacy, however, lies in the intelligent caching of popular files. To better deploy caching, a heterogeneous caching architecture is proposed that supports comprehensive cooperation. We propose three cooperative caching schemes in cellular networks, D2D networks, and cross-tier networks. Caching decisions are made by considering the content popularity, the device distribution, the transmission method, and the caching capability. Furthermore, we investigate a multi-association-based model in which a user associates with multiple caching entities to retrieve its requested content. We then present an agglomerative hierarchical clustering algorithm for setting up users' preferences and grouping them into the same clusters based on the similarity of their requests. Stochastic geometry has been used to model and analyze different coverage scenarios. Gains obtained are quantified in terms of coverage probability, cache hit probability, and delay through numerical and network simulations. Results show that the coverage probability achieved is 40% higher than the compared method. On the other hand, the cache hit probability increases to nearly 90% after clustering with the proposed method. The delay performance outperforms a popularity-based caching scheme and results in a 75% decrease in delay; however, the network's energy consumption is compromised for this purpose.

Over the last decade, deep learning has demonstrated outstanding performance in almost every application domain. Among different types of deep frameworks, convolutional neural networks (CNNs), inspired by the biological process of the visual system, can learn to extract discriminative features from raw inputs without any prior manipulation. However, efficient information circulation and the ability to explore effective new features are still two key and challenging factors for a successful deep neural network. In this thesis, we aim at presenting novel structural improvements of the CNN frameworks to enhance their effectiveness and efficiency of feature exploring and exploiting capability. To this end, first, we propose a novel residual-dense lattice network (RDL-Net), a 2-dimensional triangular lattice of convolutional units connected using residual and dense connections. RDL-Net effectively harnesses the advantages of both residual and dense aggregations without over-allocating parameters for feature re-usage. This property improves the network’s capacity to effectively and yet efficiently extract and exploit features. Furthermore, our extensive experimental investigation in processing 1D sequential speech signals shows that RDL-Nets can achieve a higher speech enhancement performance than many state-of-the-art CNN-based speech enhancement approaches. Further, we modify RDL topology to be applicable for the spatial (2D) signals. Hence, inspired by RDL-Nets innovation, we present an attention-based pyramid dilated lattice network (APDL-Net) for blind image denoising. The proposed framework employs a novel pyramid dilated convolution strategy alongside a channel-wise attention mechanism to effectively capture contextual information corresponding to different noise levels through the training of a single model. The extensive empirical studies in image denoising and JPEG artifacts suppression tasks verify the effectiveness and efficiency of the APDL architecture. We also investigate the capability of the lattice topology for hyperspectral image classification. For this purpose, we introduce a new attention-based lattice network (ALN) empowered by a unique joint spectral-spatial attention mechanism to capture spectral and spatial information effectively. The proposed ALN achieves superior accuracy and computational efficiency against state-of-the-art deep learning benchmark approaches for hyperspectral image classification. In addition to the above architectural improvements of CNNs, inspired by geographical analysis, we propose a novel channel-wise spatially autocorrelated (CSA) attention mechanism. The proposed CSA exploits the spatial relationships between feature maps channels. It also employs a unique hybrid spatial contiguity measure based on directional metrics to measure the degree of spatial closeness between feature maps effectively. Furthermore, imposing negligible learning parameters and light computational overhead to the deep model, making CSA a powerful yet efficient attention module of choice. The experimental results on large scale image classification and object detection datasets demonstrate that CSA-Nets can consistently achieve superior performance than different state-of-the-art attention-based CNNs. Besides the above architectural and attention-based advances, this research presents a simple and novel feature pooling method as gradient-based pooling (GP). This method considers the spatial gradient of the pixels within a pooling region as a key to pick the possible discriminative information. In contrast, other common pooling methods mostly rely on pixel values. The superiority of the GP over other pooling methods is proved through experiments on different benchmark image classification tasks.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Eng & Built Env
Science, Environment, Engineering and Technology
Full Text

During the last three decades, the significant increase in electricity demand, and its consequences, has appeared as a serious concern for the utility companies, but no major changes have been applied to the conventional power grid infrastructure. Recently, researchers have identified efficient control and power distribution mechanisms as the immediate challenges for conventional power grids. The next step for conventional power grid towards the Smart Grid is to provide energy efficiency management along with higher reliability via smart services, in which the application of Information and Communication Technology (ICT) is inevitable. ICT introduces powerful tools to comply with the smart grid requirements. Among various ICT properties, the telecommunication network plays a key role for providing a secure infrastructure. The two-way digital communication system provides an interaction between energy suppliers and consumers for managing, controlling and optimizing energy distribution. We can also define the smart grid as a two-way flow of energy and control information, where the electricity consumers can generate energy using green energy resources. The main objective of this thesis is to select an effective data communication infrastructure to support the smart grid services by considering a hybrid wireless and optical communication technologies. Radio-over-Fibre (RoF) networks are considered as a potential solution to provide a fast, reliable and efficient network backbone with the optical access network integration and the flexibility and mobility of the wireless network. Therefore, we adopt the integration of RoF to Passive Optical Network (PON) as a broadband access network to transmit smart grid data along with the Fiber to the Home/Building/Curb (FTTx) traffic through the shared fibre, and utilizing Wavelength Division Multiplexing (WDM). Finally, we present and analyze the simulation results for the aforementioned infrastructure based on our enhanced ROF-PON integration model.

Fault localization is the process of realizing the true source of a failure from a set of collected failure notifications. Isolating failure recovery within the network optical domain is necessary to resolve alarm storm problems. The introduction of the monitoring trail (m-trail) has been proven to deliver better performance by employing monitoring resources in a form of optical trails - a monitoring framework that generalizes all the previously reported counterparts. In this dissertation, the m-trail design is explored and a focus is given to the analysis on using m-trails with established lightpaths to achieve fault localization. This process saves network resources by reducing the number of the m-trails required for fault localization and therefore the number of wavelengths used in the network. A novel approach based on Geographic Midpoint Technique, an adapted version of the Chinese Postman’s Problem (CPP) solution and an adapted version of the Traveling Salesman’s Problem (TSP) solution algorithms is introduced. The desirable features of network architectures and the enabling of innovative technologies for delivering future millimeter-waveband (mm-WB) Radio-over-Fiber (RoF) systems for wireless services integrated in a Dense Wavelength Division Multiplexing (DWDM) is proposed in this dissertation. For the conceptual illustration, a DWDM RoF system with channel spacing of 12.5 GHz is considered. The mm-WB Radio Frequency (RF) signal is obtained at each Optical Network Unit (ONU) by simultaneously using optical heterodyning photo detection between two optical carriers. The generated RF modulated signal has a frequency of 12.5 GHz. This RoF system is easy, cost-effective, resistant to laser phase noise and also reduces maintenance needs, in principle. A revision of related RoF network proposals and experiments is also included. A number of models for Passive Optical Networks (PON)/ RoF-PON that combine both innovative and existing ideas along with a number of solutions for m-trail design problem of these models are proposed. The comparison between these models uses the expected survivability function which proved that these models are liable to be implemented in the new and existing PON/ RoF-PON systems. This dissertation is followed by recommendation of possible directions for future research in this area.

Abstract With the unprecedented growth in mobile data traffic and network densification, the emerging fifth-generation (5G) wireless network warrants a paradigm shift with respect to system design and technological enablers. In this regard, the prime motivation of this thesis is to propose an integrated access-backhaul (IAB) framework to dynamically schedule users, while efficiently providing a wireless backhaul to dense small cells and mitigating interference. In addition, joint resource allocation and interference mitigation solutions are proposed for two-hop and multi-hop self-backhauled millimeter wave (mmWave) networks. The first contribution of this thesis focuses on a multi-user two-hop relay cellular system in which a massive antenna array enabled macro base station (BS) simultaneously provides high beamforming gains to outdoor users, and wireless backhauling to outdoor small cells. Moreover, a hierarchical interference mitigation scheme is applied to efficiently mitigate cross-tier and co-tier interference. In the second contribution, a multi-hop self-backhauled mmWave communication scenario is studied whereby a joint multi-hop multi-path selection and rate allocation framework is proposed to enable Gbps data rates with reliable communications. Using reinforcement learning techniques, a dynamic and efficient re-routing solution is proposed to cope with blockage and latency constraints. Finally, a risk-sensitive learning solution is leveraged to provide high-reliability and low-latency communications. In summary, the dissertation analyses key trade-offs between (i) capacity and latency, (ii) reliability and network density. Extensive simulation results were carried out to verify the performance gains of the proposed algorithms compared to several baselines and for different network settings. Key findings show significant improvements in terms of higher data rates, lower latency, and reliable communications with some trade-offs
Tiivistelmä Liikkuvan dataliikenteen ennennäkemättömän kasvun ja verkkojen tihentymisen seurauksena pian käyttöön tulevien viidennen sukupolven (5G) langattomien verkkojen järjestelmäsuunnittelua ja teknologisten mahdollistajien käyttöä on täytynyt lähestyä kokonaan uudesta näkökulmasta. Niinpä tämän väitöstyön johtavana ajatuksena on ehdottaa integroitua verkkoon pääsyn ja runkoverkkoyhteyden muodostamismallia, jossa käyttäjät resursoidaan dynaamisesti ja samalla muodostetaan tehokkaat runkoverkkoyhteydet piensoluille. Tätä varten tutkitaan resurssiallokaation ja häiriöiden lieventämisen yhteisratkaisuja, jotka tukevat kahden tai useamman hypyn yhteyksiä ja samanaikaista runkoverkkoyhteyden luomista millimetriaaltoalueen verkoissa. Työn alkuosa keskittyy usean käyttäjän välitinavusteiseen kahden hypyn solukkoverkkoon, jossa makrotukiasemassa käytetään suurta antenniryhmää muodostamaan samanaikaisesti suuren vahvistuksen antennikeiloja käyttäjälinkeille ja langattomalle runkoyhteysosuudelle. Lisäksi sovelletaan hierarkkista häiriönvaimennusmenetelmää saman kerroksen ja kerrosten välisen häiriön tehokkaaseen vähentämiseen. Työn seuraavassa osassa arvioidaan usean hypyn runkoverkkoyhteyden muodostuksen tutkimusongelmaa millimetrialueen kommunikaatiossa kehittämällä yhdistetty menetelmä usean hypyn monipolkuvalinnalle ja tiedonsiirtoresurssien allokoinnille. Tällä tähdätään gigabittiluokan datanopeuksiin ja luotettavaan tietoliikenteeseen millimetrialueella. Vahvistavan oppimisen tekniikan avulla esitellään dynaaminen ja tehokas uudelleenreitityskonsepti toimimaan esto- ja viiverajoitusten kanssa. Lopuksi hyödynnetään riskisensitiivistä oppimista ja antennidiversiteettitekniikoita suuren luotettavuuden ja pienen latenssin saavuttamiseksi millimetrialueen tiedonsiirrossa. Näiden avulla analysoidaan kaupankäyntiä esimerkiksi (i) kapasiteetin ja latenssin sekä (ii) luotettavuuden ja verkon tiheyden/kuormituksen välillä. Mittavien suoritettujen simulointien avulla osoitetaan ehdotettujen algoritmien suorituskykyedut suhteessa tunnettuihin verrokkeihin useissa eri skenaarioissa. Tulosten perusteella saavutetaan merkittäviä kustannussäästöjä infrastruktuurin ja runkoverkon osalta sekä päästään suuriin datanopeuksiin ja parannuksiin pienen latenssin luotettavassa tietoliikenteessä

Dans cette thèse, nous développons des algorithmes à haute performance pour certains calculs impliquant des tenseurs denses et des matrices éparses. Nous abordons les opérations du noyau qui sont utiles pour les tâches d'apprentissage de la machine, telles que l'inférence avec les réseaux neuronaux profonds. Nous développons des structures de données et des techniques pour réduire l'utilisation de la mémoire, pour améliorer la localisation des données et donc pour améliorer la réutilisation du cache des opérations du noyau. Nous concevons des algorithmes parallèles à mémoire séquentielle et à mémoire partagée.Dans la première partie de la thèse, nous nous concentrons sur les noyaux tenseurs denses. Les noyaux tenseurs comprennent la multiplication tenseur-vecteur (TVM), la multiplication tenseur-matrice (TMM) et la multiplication tenseur-tendeur (TTM). Parmi ceux-ci, la MVT est la plus liée à la largeur de bande et constitue un élément de base pour de nombreux algorithmes. Nous proposons une nouvelle structure de données qui stocke le tenseur sous forme de blocs, qui sont ordonnés en utilisant la courbe de remplissage de l'espace connue sous le nom de courbe de Morton (ou courbe en Z). L'idée clé consiste à diviser le tenseur en blocs suffisamment petits pour tenir dans le cache et à les stocker selon l'ordre de Morton, tout en conservant un ordre simple et multidimensionnel sur les éléments individuels qui les composent. Ainsi, des routines BLAS haute performance peuvent être utilisées comme micro-noyaux pour chaque bloc. Les résultats démontrent non seulement que l'approche proposée est plus performante que les variantes de pointe jusqu'à 18%, mais aussi que l'approche proposée induit 71% de moins d'écart-type d'échantillon pour le MVT dans les différents modes possibles. Enfin, nous étudions des algorithmes de mémoire partagée parallèles pour la MVT qui utilisent la structure de données proposée. Nos résultats sur un maximum de 8 systèmes de prises montrent une performance presque maximale pour l'algorithme proposé pour les tenseurs à 2, 3, 4 et 5 dimensions.Dans la deuxième partie de la thèse, nous explorons les calculs épars dans les réseaux de neurones en nous concentrant sur le problème d'inférence profonde épars à haute performance. L'inférence sparse DNN est la tâche d'utiliser les réseaux sparse DNN pour classifier un lot d'éléments de données formant, dans notre cas, une matrice de caractéristiques sparse. La performance de l'inférence clairsemée dépend de la parallélisation efficace de la matrice clairsemée - la multiplication matricielle clairsemée (SpGEMM) répétée pour chaque couche dans la fonction d'inférence. Nous introduisons ensuite l'inférence modèle-parallèle, qui utilise un partitionnement bidimensionnel des matrices de poids obtenues à l'aide du logiciel de partitionnement des hypergraphes. Enfin, nous introduisons les algorithmes de tuilage modèle-parallèle et de tuilage hybride, qui augmentent la réutilisation du cache entre les couches, et utilisent un module de synchronisation faible pour cacher le déséquilibre de charge et les coûts de synchronisation. Nous évaluons nos techniques sur les données du grand réseau du IEEE HPEC 2019 Graph Challenge sur les systèmes à mémoire partagée et nous rapportons jusqu'à 2x l'accélération par rapport à la ligne de base
In this thesis, we develop high performance algorithms for certain computations involving dense tensors and sparse matrices. We address kernel operations that are useful for machine learning tasks, such as inference with deep neural networks (DNNs). We develop data structures and techniques to reduce memory use, to improve data locality and hence to improve cache reuse of the kernel operations. We design both sequential and shared-memory parallel algorithms. In the first part of the thesis we focus on dense tensors kernels. Tensor kernels include the tensor--vector multiplication (TVM), tensor--matrix multiplication (TMM), and tensor--tensor multiplication (TTM). Among these, TVM is the most bandwidth-bound and constitutes a building block for many algorithms. We focus on this operation and develop a data structure and sequential and parallel algorithms for it. We propose a novel data structure which stores the tensor as blocks, which are ordered using the space-filling curve known as the Morton curve (or Z-curve). The key idea consists of dividing the tensor into blocks small enough to fit cache, and storing them according to the Morton order, while keeping a simple, multi-dimensional order on the individual elements within them. Thus, high performance BLAS routines can be used as microkernels for each block. We evaluate our techniques on a set of experiments. The results not only demonstrate superior performance of the proposed approach over the state-of-the-art variants by up to 18%, but also show that the proposed approach induces 71% less sample standard deviation for the TVM across the d possible modes. Finally, we show that our data structure naturally expands to other tensor kernels by demonstrating that it yields up to 38% higher performance for the higher-order power method. Finally, we investigate shared-memory parallel TVM algorithms which use the proposed data structure. Several alternative parallel algorithms were characterized theoretically and implemented using OpenMP to compare them experimentally. Our results on up to 8 socket systems show near peak performance for the proposed algorithm for 2, 3, 4, and 5-dimensional tensors. In the second part of the thesis, we explore the sparse computations in neural networks focusing on the high-performance sparse deep inference problem. The sparse DNN inference is the task of using sparse DNN networks to classify a batch of data elements forming, in our case, a sparse feature matrix. The performance of sparse inference hinges on efficient parallelization of the sparse matrix--sparse matrix multiplication (SpGEMM) repeated for each layer in the inference function. We first characterize efficient sequential SpGEMM algorithms for our use case. We then introduce the model-parallel inference, which uses a two-dimensional partitioning of the weight matrices obtained using the hypergraph partitioning software. The model-parallel variant uses barriers to synchronize at layers. Finally, we introduce tiling model-parallel and tiling hybrid algorithms, which increase cache reuse between the layers, and use a weak synchronization module to hide load imbalance and synchronization costs. We evaluate our techniques on the large network data from the IEEE HPEC 2019 Graph Challenge on shared-memory systems and report up to 2x times speed-up versus the baseline

Grâce aux avancées technologiques dans le domaine des réseaux cellulaires et des équipements mobiles, le nombre d'applications multimédia à haut débit dans les réseaux mobiles ne cesse d'augmenter. On prévoit que le trafic de données dans les réseaux mobiles en 2017 sera 13 fois plus important que celui en 2012. Pour satisfaire aux besoins des équipements mobiles, de nouvelles approches pour la gestion des ressources radio et des puissances de transmission sont requises.Dans le cadre de cette thèse, on s'intéresse à proposer des solutions pour remédier aux problèmes des interférences intercellulaires dans les réseaux mobiles de dernière génération. Nous enquêtons d'une manière exhaustive les différentes techniques de coordination des interférences intercellulaires existantes. Ces techniques sont qualitativement comparées, puis classées selon le taux de coopération requis entre les différentes stations de base, mais aussi selon leurs principes de fonctionnement. Nous abordons également le problème multicellulaire d'allocation des ressources et des puissances de transmission d'une manière centralisée. Nous formulons ce problème d'optimisation centralisé, puis nous le décomposons en deux sous-problèmes indépendants : l'allocation de ressources et l'allocation des puissances de transmission. De plus, une approche distribuée basée sur la théorie des jeux est proposée pour l'allocation des puissances de transmission. Les techniques centralisées de minimisation des interférences intercellulaires offrent la solution optimale au prix d'une grande charge de signalisation. Par contre, les solutions décentralisées réduisent le trafic de signalisation sans garantir l'optimalité de la solution obtenue. Nous proposons ensuite une heuristique de contrôle de puissance qui modifie localement l'allocation des puissances de transmission de manière à éviter le gaspillage d'énergie et pour réduire les interférences ressenties par les utilisateurs des stations de base voisines. Nous proposons également une technique autonome qui gère la distribution des ressources radio entre les différentes zones de chaque cellule. Cette technique répond aux besoins des utilisateurs dans chaque zone en adaptant la distribution des ressources d'une manière dynamique. Nous abordons aussi le compromis entre les techniques de gestion d'interférences intercellulaires centralisées et décentralisées. Nous proposons une approche hybride où l'allocation des ressources radio et des puissances de transmission est faite d'une manière coopérative entre les différentes cellules. Dans un premier lieu, les cellules voisines collaborent afin d'ajuster les puissances de transmission allouées aux ressources radio. Ensuite, la distribution des ressources entre les différentes zones de chaque cellule est modifiée localement, selon les besoins des utilisateurs dans chaque zone
The exponentially increasing demand for mobile broadband communications have led to the dense deployment of cellular networks with aggressive frequency reuse patterns. The future Fifth Generation (5G) networks are expected to overcome capacity and throughput challenges by adopting a multi-tier architecture where several low-power Base Stations (BSs) are deployed within the coverage area of the macro cell. However, Inter-Cell Interference (ICI) caused by the simultaneous usage of the same spectrum in different cells, creates severe problems. ICI reduces system throughput and network capacity, and has a negative impact on cell-edge User Equipment (UE) performance. Therefore, Inter-Cell Interference Coordination (ICIC) techniques are required to mitigate the impact of ICI on system performance. In this thesis, we address the resource and power allocation problem in multiuser Orthogonal Frequency Division Multiple Access (OFDMA) networks such as LTE/LTE-A networks and dense small cell networks. We start by overviewing the state-of-the-art schemes, and provide an exhaustive classification of the existing ICIC approaches. This qualitative classification is followed by a quantitative investigation of several interference mitigation techniques. Then, we formulate a centralized multi-cell joint resource and power allocation problem, and prove that this problem is separable into two independent convex optimization problems. The objective function of the formulated problem consists in maximizing system throughput while guaranteeing throughput fairness between UEs. ICI is taken into account, and resource and power allocation is managed accordingly in a centralized manner. Furthermore, we introduce a decentralized game-theoretical method to solve the power allocation problem without the need to exchange signaling messages between the different cells. We also propose a decentralized heuristic power control algorithm based on the received Channel Quality Indication (CQI) feedbacks. The intuition behind this algorithm is to avoid power wastage for UEs that are close to the serving cell, and reducing ICI for UEs in the neighboring cells. An autonomous ICIC scheme that aims at satisfying throughput demands in each cell zone is also introduced. The obtained results show that this technique improves UE throughput fairness, and it reduces the percentage of unsatisfied UEs without generating additional signaling messages. Lastly, we provide a hybrid ICIC scheme as a compromise between the centralized and the decentralized approaches. For a cluster of adjacent cells, resource and power allocation decisions are made in a collaborative manner. First, the transmission power is adjusted after receiving the necessary information from the neighboring cells. Second, resource allocation between cell zones is locally modified, according to throughput demands in each zone

Dissertação de Mestrado em Engenharia Informática apresentada à Faculdade de Ciências e Tecnologia
Nos últimos anos o número de redes de Internet das Coisas tem aumentado. Para suportar redes de 5ª Geração (5G) redes altamente densas de larga escala vão ter de ser implementadas. Essas redes vão conter um número enorme de sensores a pilhas, detectando evento e reenviando mensagens em topologias dinâmicas (estrela, malha, ad-hoc) Estas redes altamente densas podem levar um nó a esgotar os seu recursos rapidamente. Portanto essas redes terão de ser tão eficientes quanto possível para operar durante o tempo que for necessário, mantendo as comunicações fiáveis.Este trabalho examina e apresenta um estado-da-arte sobre metodologias de medição de energia e de agregação de dados em redes de baixa potência. Medir o consumo de energia de múltiplos sensores é uma tarefa complexa. Este trabalho apresenta algumas das técnicas usadas e opta por um software para medir o consumo. Este trabalho foca-se também em agregação de dados dentro da rede. Esta agregação é feita em cada salto.A parte crítica deste trabalho é focada no desenvolvimento de um ambiente de testes. Este ambiente de testes consiste em várias placas que comunicam entre si e com um gateway. O objetivo do ambiente de testes é medir o consumo de energia em vários cenários.Com estes desafios em mente, este trabalho apresenta uma abordagem multi-camada para realizar a agregação de dados. O objetivo principal da agregação é reduzir o consumo de energia. Este método é baseado na criação de grupos de nós com configurações semelhantes, aproveitando a semelhança desses dados. O mecanismo final foi capaz de atingir até 9.17% de melhoria no consumo de energia realizando agregação.
In the last few years the number of Internet of Things (IOT) networks has been increasing. In order to support Fifth Generation (5G), large-scale highly- dense networks will have to be deployed. Those networks will contain a massive number of low power, battery operated sensors, sensing and forwarding messages in dynamic topologies (star, mesh, ad-hoc). These highly dense networks can cause rapid exhaustion of a node’s resources. As such they have to be as efficient as possible to operate as long as they are needed, while achieving reliable communications. This work presents and examines state of the art mechanisms for energy measurements and data aggregation in Low power and Lossy network (LLN). Measuring the energy consumption of multiple sensor nodes is a complex task. This work presents some of the techniques used and opts by a software approach to obtain that metric. This work focuses on in-network data aggregation. The data aggregation is performed at every hop. The core part of this work focuses on the development of a testbed environment. This environment consists of several physical boards communicating with each other and a gateway. The main focus of the testbed is measuring the energy consumption across different scenarios. With these challenges in mind, this work presents a cross-layer approach to data aggregation. The main objective of the aggregation is to reduce the power consumption. The method is based on the creation of groups of nodes with similar properties, leveraging the similarity of the exchanged data. The final mechanism is capable of achieving up to 9.17% in energy savings when performing aggregation.