Добірка наукової літератури з теми "004.25:004.62"

Дисертація на здобуття наукового ступеня кандидата технічних наук за спеціальністю 05.13.05 – комп’ютерні системи та компоненти. – Національний технічний університет "Харківський політехнічний інститут", Харків, 2019. Дисертаційна робота присвячена вирішенню науково-практичної задачі, яка полягає в розробленні та дослідженні моделей та структур даних на основі асоціативної пам’яті для комп’ютерних систем проектування структур технологічних процесів. Метою дисертаційної роботи є розроблення структури даних на основі асоціативної пам’яті для комп’ютерних систем проектування структур технологічних процесів. Об’єкт дослідження – процеси обробки інформації з використанням асоціативної пам’яті. Предмет дослідження – моделі структур даних на основі асоціативної пам’яті для комп’ютерних систем проектування структур технологічних процесів. У вступі обґрунтовано актуальність теми дисертаційної роботи; сформульовано мету та задачі дослідження; наведено зв’язок дисертації з науковими темами та програмами; визначено об’єкт та предмет дослідження; висвітлені положення, які визначають практичне значення отриманих результатів та їх наукову новизну; висвітлено особистий внесок здобувача в одержаних результатах, їх апробацію та публікації. У першому розділі проведено аналіз переваг та недоліків сучасних методів зберігання та обробки інформації при проектуванні структур технологічних процесів, який показав, що існуючі методи на основі стандартного зберігання не забезпечують необхідної швидкодії для реалізації інтелектуальних інтерфейсів. При створенні одиничних технологічних процесів з використанням баз даних, час економиться на те, що текст переходів в тій чи іншій мірі вже присутній в базі даних, але вибір стратегії обробки, обладнання та інструменту, залишається в руках технологів. Всі необхідні дані повинні зберігатися в базі і бути доступними при необхідності отримання інформації. Створити таку базу даних дозволяють структури даних на основі асоціативної пам’яті, які можуть зберігати великий обсяг інформації та дають можливість коригувати її в процесі експлуатації. Тому виникає необхідність розробки такої асоціативної пам’яті, що дозволить зберігати та обробляти велику кількість формалізованої інформації, яка на даний час зберігається у вигляді базових структур (списки, черги, дерева та ін.). Що, в свою чергу, дозволить будувати ТП для складних, високоточних виробів, з можливістю коригування даних в процесі побудови ТП та асоціювати з однією деталлю декілька технологічних процесів. В роботі пропонується для вирішення цієї задачі виконати розроблення та дослідження нейронних мереж асоціативної пам’яті, моделей та структур даних для комп’ютерної системи проектування технологічних процесів. Другий розділ присвячено розробці нових нейронних мереж на основі двонаправленої асоціативної пам’яті. Запропоновано нейронну мережу N-направленої асоціативної пам'яті, що складається з N шарів нейронів та дозволяє зберігати множинні асоціації. Запропоновано нейронні мережі асоціативної пам’яті, що здатні відновлювати по вхідній інформації безліч асоціативних вхідних зображень, по множині вхідних даних один вихідний елемент та по комбінації вхідних елементів множину вихідних. Тобто розроблена асоціативна пам’ять 1→ N , N →1 та 1 2 N → N . Запропоновано багатошарову асоціативну пам'ять з керуючими нейронами, що дозволяє будувати ланцюги асоціативних даних з можливістю коригування в процесі функціонування мережі. Розроблені нейронні мережі на основі асоціативної пам’яті, які можуть бути використані при проектуванні складних нейронних мереж деревовидної структури, що можуть використовуватися при побудові системи зберігання та обробки даних в інформаційних системах. У третьому розділі розглядається розроблення нових видів асоціативної пам’яті, які дозволяють зберігати інформацію на основі базових структур та комбінувати їх для ефективної обробки інформації. При побудові складної системи зберігання інформації в якості елементів пам'яті використовуються не окремі осередки, а їх об'єднання, що зберігають такі структури як списки, масиви, черги і т.п. Кожна з них являє собою структурований елемент, завдяки якому забезпечується пошук і обробка даних, що зберігаються в ньому. Багаторівневе комбінування цих елементів, кожен з яких можна представити як деяку множину елементів асоціативної пам'яті, є перспективним для реалізації інтелектуальних інтерфейсів, логічного висновку, обробки зображень, розпізнавання образів. У четвертому розділі розглядаються моделі структур даних на основі нейронних мереж асоціативної пам’яті для проектування структур технологічних процесів. Запропоновано структури даних на основі асоціативної пам’яті для зберігання інформації, що необхідна для проектування структури технологічного процесу. Запропоновано ієрархічну нейронну мережу на основі асоціативної пам’яті для вибору обладнання для однієї операції технологічного процесу. Результати роботи впроваджені: – в державному підприємстві "Південний державний проектно- конструкторський та науково-дослідний інститут авіаційної промисловості" (м. Харків) при виготовленні високоточних виробів; – в навчальному процесі НТУ "ХПІ" на кафедрі обчислювальної техніки та програмування при викладанні учбових дисциплін: "Основи нейрокомп’ютингу", "Системи штучного інтелекту", а також в курсових, дипломних та наукових роботах студентів спеціальності 123 "Комп’ютерна інженерія".
There is the dissertation of the obtaining the scientific degree of the technical sciences candidate in specialty 05.13.05 – computer systems and components – National Technical University "Kharkiv Polytechnic Institute", Kharkiv, 2019. The dissertation solves the scientific and practical problem of development and researching models data structures based on associative memory for computer systems of technological processes structure design. The purpose of the dissertation is the development and researching of data structures based on associative memory for computer systems of technological processes structure design. The object of study is the processes of processing information using associative memory. The subject of study is the models of data structures based on associative memory for computer systems of technological processes structure design. The introduction substantiates the relevance of the dissertation work topic; the purpose and tasks of research are formulated; the connection of the dissertation with scientific themes and programs is given; the object and the subject of the research are determined; the provisions, that determine the practical value of the obtained results and their scientific novelty, are highlighted; the personal contribution of the applicant in the obtained results, their approbation and publication are highlighted. In the first section the analysis of the advantages and disadvantages of current methods of storing and processing information for the design of structures processes, which showed that existing methods based on standard storage do not provide the performance required for the implementation of intelligent interfaces. When creating single technological processes using databases, time is saved by the fact that the transition text is already present in the database to some extent, but the choice of processing strategy, equipment and tools remains in the hands of technologists. All necessary data should be stored in the database and be accessible when necessary to obtain information. Such a database is made possible by associative memory data structures, which can store a large amount of information and allow it to be corrected during operation. Therefore, there is a need to develop associative memory that will allow the storage and processing of a large amount of formalized information, which is currently stored as basic structures (lists, queues, trees, etc.). This, in turn, will allow the construction of TP for complex, high-precision products, with the ability to correct data in the process of TP construction and associate several technological processes with one detail. In order to solve this problem, the paper proposes to develop and research neural networks of associative memory, models and data structures for the computer system of technological process design. The second section is devoted to the development of new neural networks based on bidirectional associative memory. A neural network of N-directed associative memory, consisting of N layers of neurons and allowing multiple associations to be stored, is proposed. Associative memory neural networks have been proposed that are capable of restoring a plurality of associative input images by input, by multiple inputs, by one output element, and by a combination of input elements by multiple outputs. That is, the associative memory 1→ N , N →1 and 1 2 N → N were developed. Multilayer associative memory with control neurons is proposed, which allows the construction of associative data circuits with the possibility of correction during the operation of the network. Developed neural networks based on associative memory that can be used in the design of complex neural networks of tree structure that can be used in the construction of storage and data processing systems in information systems. The second section is devoted to the development of new types of associative memory that allow us to store information based on underlying structures and combine them for efficient information processing. When constructing a complex system of information storage, the memory elements are used not by individual cells, but by their associations, which store such structures as lists, arrays, queues, etc. Each of them is a structured element by which search and processing of the data stored in it is provided. The multilevel combination of these elements, each of which can be represented as some of the elements of associative memory, is promising for the implementation of intelligent interfaces, logical inference, image processing, pattern recognition. The fourth section discusses the model data structures based on neural networks, associative memory structures to design processes. Data structures based on associative memory are proposed for storing the information needed to design the process structure. A hierarchical neural network based on associative memory is proposed to select equipment for a single process operation. The results of the dissertation work are introduced: – in the state enterprise "State project and scientific aviation industry research institute" in the manufacture of high-precision products; – in the educational process of the NTU "KhPI" at the department of computer engineering and programming during the teaching of the academic disciplines: "Fundamentals of Neurocomputing", "Artificial intelligence systems", and also in courses, thesis and scientific works of students of the specialty 123 "Computer engineering".

The ever growing Internet Protocol (IP) traffic driven by both consumer and business applications pose a fundamental challenge for network operators. Optical networks offering terabyte capacity and advanced switching capability is at the forefront of transport technology. As optical networks and data traffic proliferate there is a growing need to scale cost effectively and allow for multiple switching technologies. Addressing this, the G.709 Optical Transport Network (OTN) combined with the intelligence of the Generalised Multi Protocol Label Switching (GMPLS) control plane is identified to reduce transport costs and deliver enhanced network and performance management functions. Connections are established using GMPLS suite of protocols namely routing, signaling and the Path Computation Element (PCE). The conventional wavelength resource reservation mechanism in the Optical Channel (OCh) layer is stud- ied and shortcomings identified. Two distributed wavelength resource reservation algorithms that improve end-to-end wavelength allocation are proposed and offer significant network blocking performance improvement. Further, the Resource Reservation Protocol with Traffic Engineering (RSVP- TE) / PCE wavelength reservation is extended to the newly defined Optical channel Data Unit (ODU) digital layer to provision at sub-wavelength granularity. Network blocking performance, rough estimates of unit cost and wavelength utilisation in comparison with OCh layer are analysed showing improved network blocking performance and wavelength utilisation at the expense of increased cost unit. Also at higher transmission bit rates of 40Gbps and beyond, it is inefficient to provision light paths without taking into consideration the optical performance requirements of the client signal. To demonstrate Impairment-Aware (lA) wavelength resource reservation, the conventional RSVP- TE and best performing proposed algorithm are adapted for 10/ 40 Gbps client signals and network blocking performance is observed. Further, hardware-accelerated control plane functions to overcome challenges posed by overhead processing for short-lived connections in dynamic high-speed networks is discussed.

Electronic services are emerging as the de-facto enabler of interaction interoperability across organization boundaries. Cross-organizational interactions are often "choreographed", i.e. specified by a messaging protocol from a global point of view independent of the local view of each interacting organization. Local requirements motivating an interaction as well as the global contextual requirements governing the interaction inevitably evolve over time, requiring adaptation of the corresponding interaction protocol. Adaptation of an interaction protocol must ensure the satisfaction of both sets of interaction requirements while maintaining consistency between the global view and the local views of an interaction specification. Such adaptation is not possible with the current state-of-the-art representations of choreographed interactions, as they capture only operational messaging specifications detached from both local organizational requirements as well as global contextual requirements. This thesis presents three novel contributions that tackle adaptation of choreographed interaction protocols: an automated technique for deriving an interaction protocol from requirements, a formalization of consistency between local and global views, and a framework for guiding the adaptation of a choreographed interaction. A choreographed interaction is specified using models of organizational requirements motivating the interaction. We employ the formal semantics embedded in requirements models to automatically derive an interaction protocol. We propose a framework for relating the global and local views of interaction specification and maintaining consistency between them. We develop a metamodel for interaction specification, from which we enumerate adaptation operations. We build a catalogue that provides guidance on performing each operation and propagating changes between the global and local views. These contributions are evaluated using examples from the literature as well as a real-world case study.

The emergence of Bluetooth as a wireless network solution assists in bringing together multiple technologies in different sectors and provides rapid interconnections to form a network paradigm. Typically, up to 8 Bluetooth devices can form a centralized network, called a piconet, controlled by a master node, which allocates transmission slots to all other nodes (slaves) in the piconet. However, the structure of inter-piconet connection: called the scatternet is not defined in the Bluetooth Specification. To develop a new scatternet structure, many challenges such as topology formation, intra and inter-piconet scheduling and packet routing are considered. The thesis addresses these critical issues based on the scatternet formation using Bluetooth. The scatternet, presented in this thesis, employs a tree hierarchy structure formation with a Leader root of three hierarchies. Within the scatternet, the new concept exploits clock and frequency synchronization for all new piconets creation. This synchronization prevents interferences and the need for guard time while switching piconet. Thus enabling a device to switch from one piconet to another at every slot. Furthermore, an innovative intra-piconet design is proposed to improve QoS within Bluetooth piconet. By exploiting the device queue status, the scheme defines a predictable polled sequence, and an adaptive traffic allocation. This offers a better fairness, and a significant power reduction when compared to the conventional Round Robin scheme. Moreover, with the perfect scatternet synchronization, the devices switch to other piconets to transmit data and, within one slot time, return to the initial piconet before the next predicted poll time exchange occurs. This considerably improves the traffic data transfer, especially for a significant number of devices present within the area. In addition, a new routing process is developed in this thesis, which facilitates communication within the scatternet. From the scatternet tree hierarchy position, a new addressing node routing is proposed to keep the overhead network low, and guarantees that any packet forwarded reaches its destination. The performance of the new scatternet is evaluated for each scatternet phase, through a Matlab simulation program, and the significant improvement of Bluetooth QoS achieved through the proposed approach is fully demonstrated. This thesis also presents a primary implementation of the scatternet concept using a dedicated Bluetooth hardware system.

This thesis reports a series of five experiments which were designed to assess two research objectives; 1) to ascertain whether CMC improves episodic and semantic collaborative recall and 2) to challenge the existing assumptions regarding semantic recall by establishing if "the nominal group effect" prevails in semantic recall. In each experiment, nominal group recall was compared with different types of collaborative recall, which included face-to-face collaboration and two types of computer mediated communication (CMC), namely synchronous (parallel) and asynchronous (serial/cyclic). Parallel communication comprised the simultaneous communication of group members in an online environment who had complete freedom over when they made contributions and attended/responded to other group members. Serial/cyclic communication employed a taking procedure where group members were exposed to each others' answers, but only one group member could contribute at a time. Two types of turn-taking procedure were examined; serial was a once-through approach where each group member got one chance to contribute (Experiment 1) and cyclic was a round-robin protocol where group members continually took turns to contribute for the duration of the recall trial (Experiments 3- 5). In Experiment 1, participants generated items from semantic categories as nominal, face-to-face, parallel and serial triads, i.e. groups. Results failed to provide evidence of the nominal group effect and there was no evidence to suggest a benefit from collaborating using one medium over another. Experiment 2 examined episodic and semantic recall in nominal, face-to-face and parallel collaboration. For episodic retrieval, participants were required to memorise categorised word lists and for semantic retrieval, participants were required to generate words beginning with specified orthographic digraphs i.e. 'br', 'he', 'po'. Results demonstrated a nominal group effect throughout episodic and semantic retrieval. Experiment 3 examined the effects of CMC when task complexity was increased to encompass the generation of words from a fixed set of letters in a Scrabble task. Once again, nominal groups generated the highest number of items, followed by parallel, face-to-face and cyclic groups. Experiments 4 and 5 extended the recent work of Finlay, Hitch and Meudell (200 1) and Blumen arid Rajaram (2008), who found that when collaborative retrieval was followed by individual retrieval, individuals were able to benefit from prior exposure, despite their initial losses acquired collaboratively. Results demonstrated improved individual episodic recall following face-to-face, parallel and cyclic collaboration, but there was no benefit for semantic retrieval.

The initial rapid expansion of the Internet, in terms of complexity and number of hosts, was followed by an increased interest in its overall parameters and the quality the network offers. This growth has led, in the first instance, to extensive research in the area of network monitoring, in order to better understand the characteristics of the current Internet. In parallel, studies were made in the area of protocol performance modelling, aiming to estimate the performance of various Internet applications. A key goal of this research project was the analysis of current Internet traffic performance from a dual perspective: monitoring and prediction. In order to achieve this, the study has three main phases. It starts by describing the relationship between data transfer performance and network conditions, a relationship that proves to be critical when studying application performance. The next phase proposes a novel architecture of inferring network conditions and transfer parameters using captured traffic analysis. The final phase describes a novel alternative to current TCP (Transmission Control Protocol) models, which provides the relationship between network, data transfer, and client characteristics on one side, and the resulting TCP performance on the other, while accounting for the features of current Internet transfers. The proposed inference analysis method for network and transfer parameters uses online nonintrusive monitoring of captured traffic from a single point. This technique overcomes limitations of prior approaches that are typically geared towards intrusive and/or dual-point offline analysis. The method includes several novel aspects, such as TCP timestamp analysis, which allows bottleneck bandwidth inference and more accurate receiver-based parameter measurement, which are not possible using traditional acknowledgment-based inference. The the results of the traffic analysis determine the location of the eventual degradations in network conditions relative to the position of the monitoring point. The proposed monitoring framework infers the performance parameters of network paths conditions transited by the analysed traffic, subject to the position of the monitoring point, and it can be used as a starting point in pro-active network management. The TCP performance prediction model is based on the observation that current, potentially unknown, TCP implementations, as well as connection characteristics, are too complex for a mathematical model. The model proposed in this thesis uses an artificial intelligence-based analysis method to establish the relationship between the parameters that influence the evolution of the TCP transfers and the resulting performance of those transfers. Based on preliminary tests of classification and function approximation algorithms, a neural network analysis approach was preferred due to its prediction accuracy. Both the monitoring method and the prediction model are validated using a combination of traffic traces, ranging from synthetic transfers/environments, produced using a network simulator/emulator, to traces produced using a script-based, controlled client and uncontrolled traces, both using real Internet traffic. The validation tests indicate that the proposed approaches provide better accuracy in terms of inferring network conditions and predicting transfer performance in comparison with previous methods. The non-intrusive analysis of the real network traces provides comprehensive information on the current Internet characteristics, indicating low-loss, low-delay, and high-bottleneck bandwidth conditions for the majority of the studied paths. Overall, this study provides a method for inferring the characteristics of Internet paths based on traffic analysis, an efficient methodology for predicting TCP transfer performance, and a firm basis for future research in the areas of traffic analysis and performance modelling.