【作者】 金鑫；

【导师】 刘贤德；

【作者基本信息】 华中科技大学 ， 物理电子学， 2005， 博士

【摘要】 随着信息技术、计算机技术、网络技术和控制技术的进步,智能建筑得到了大力的推广和发展,信息时代构筑智能空间已逐渐成为现实。目前,智能空间的构筑仅仅局限于控制系统的网络互联,而不同系统之间的信息与资源共享率极为低下。智能空间应该能够有机地互联各个分离的智能建筑,以实现控制网络与系统、数据网络与系统、语音网络与系统以及视频网络与系统等等的无缝融合。而基于Internet 的互联则是实现智能空间的关键技术之一,但是IP 网络仅能为各种业务提供尽力传送的服务,而并不保证提供业务要求的服务质量。本文围绕基于IP 网络的智能空间建模与模型网络的服务质量单播路由的相关理论与算法开展了研究工作,在基于IP 网络的智能空间建模、消除模型网络中不精确状态信息对路由的影响、模型网络的各种单播路由算法,如分布式路由算法、源路由算法、群集智能路由算法以及流量工程在算法中的应用等等方面进行了细致的研究。(1) 以现有的计算机网络技术、语音压缩与传输技术、视频压缩与传输技术和楼宇自动控制网络技术为研究背景,构造了一种基于IP 网络的智能空间模型。该模型实现了智能空间中的控制子空间、数据子空间、语音子空间以及视频子空间等等的互联与集成,具有强大的信息与资源共享功能。(2) 根据智能空间网络不精确状态信息产生的原因,提出了一种新颖的思路,用于解决不精确信息对路由算法带来不利影响的问题。该思路不需要对状态信息进行极值分布的假定,而是利用已有的状态信息选择概率上受不精确信息影响最小的、满足服务质量要求的路径,从而有效地减轻智能空间中网络信息不精确性对算法的影响。(3) 提出了一种基于优化路径的扩展Dijkstra 路由算法,用于解决智能空间中不精确信息下的多限制路径问题。算法采用组合参数进行搜索,降低了信息不精确对算法的影响,并优化了资源的利用;通过限制节点优化路径的数目,降低了算法的复杂度。仿真说明了算法具有简单、高效、可扩展、收敛快等特点,可以在一定程度上减弱状态信息不精确对算法的影响,并为智能空间中的业务提供多限制路径路由能力。(4) 提出了一种基于优化路径的有限选择洪泛单播路由算法,用于解决智能空间更多还原

【Abstract】 With the development of information technology, computer technology, network technology and control technology, intelligent buildings are more and more popular, and intelligent spaces are being realized. Now the intelligent spaces are only to interlink the automation and control systems of the buildings, and they don’t share their information and resources between different systems. They should be able to interlink separate intelligent building, and provide the perfect integration of control network and system, data network and system, voice network and system and video network and system. Interlinkage via Internet is a key technology to realize intelligent space, but IP network could only provide the "Best Effort" service, and could not ensure the quality of service. In this dissertation, the author presents the academic and algorithmic researches on intelligent space model based on IP network and quality of service unicast routing in model network. And the researches involve intelligent space model based on IP network, eliminating the influence of the imprecise state information to routing algorithms, various unicast routing algorithms of model network, like distributed routing algorithm, source routing algorithm and swarm intelligence routing algorithm, utilization of traffic engineering, and so on. (1) Constructs an intelligent space model based on IP network, according to the computer and network technology, voice compression and transmission technology, video compression and transmission technology, and building automation and control technology. The model could realize the perfect integration of control network and system, data network and system, voice network and system and video network and system, and provide powerful information and resource share. (2) Proposes a novel method to solve the influence of imprecise state information in intelligent space based on the cause of imprecision. It doesn’t need any assumption for probability distribution or extremum of state information, and use existing information to choose an appropriate path meeting constraints of quality of service, which is least affected in probability by imprecise information. It can effectively weaken the influence to routing algorithms caused by network state information imprecision in intelligent space. (3) Proposes an extended Dijkstra unicast routing algorithm based on non-dominated paths for solving the multi-constrained path problem with imprecise information in intelligent space. It can weaken the influence of state information imprecision and efficiently utilize resource by searching paths with a combined main parameter of all metric. And it can reduce the computational complexity by limiting the counts of non-dominated paths kept at network nodes. The computerized simulations show that the algorithm is simple, efficient, scaleable and quickly convergent. It can weaken the influence of state information imprecision to a certain extend, and provide multi-constrained routing choices for the operations and applications of intelligent space. (4) Proposes a limited selective flooding unicast routing algorithm based on non-dominated paths for solving the multi-constrained path problem with imprecise information in intelligent space. It can reduce the time complexity and the space complexity by limiting the counts of non-dominated paths kept at network nodes and restricting the usage times of all network links. The computerized simulations show that the algorithm is simple, efficient, scaleable and quickly convergent. It can effectively weaken the influence of state information imprecision, and provide multi-constrained routing choices for the operations and applications of intelligent space. (5) Proposes a distributed unicast routing framework with extensive applicability, which is designed to solve NP-complete routing problems with imprecise information in intelligent space. Based on the framework, the typical NP-complete routing problems of intelligent space are analyzed and solved. The computerized simulations show that the framework is simple, scaleable and quickly convergent. It can effectively weaken the influence of state information imprecision, effectually solve the NP-complete routing problem, and provide routing choices with quality of service for various operations and applications of intelligent space.(6) Proposes an improved artificial ant colony algorithm for solving the problem that the capacities of paths are limited. It modifies the path selection rule by combining traffic engineering and can provide load balance for multi paths. And it can strengthen load balance of networks if it is used by routing algorithm. (7) Proposes a swarm intelligence unicast routing strategy based on local network state information. It doesn’t need the network update of state information, while provides a same routing table for all applications by utilizing mobile agents which are unrelated with applications and perform the improved artificial ant colony algorithm. And it searches an appropriate path for an application call by utilizing mobile agents which are corresponding to the call and contain appointed quality of service parameters. It is extensively appliable for various quality of service routing problems of intelligent space, is able to provide routing choices of quality of service for the operations and applications, and can ensure good scaleable for new future applications. (8) Constructs an extended Dijkstra unicast routing algorithm based on traffic engineering for solving the bandwidth constrained delay constrained path problem, and research the influence of traffic engineering to routing algorithms. The computerized simulations show that the algorithm is simple and efficient, and can increase resource usage ratio and efficiently reduce the call bock ratio. It can adapt networks whose bandwidths of links are not same, and effectively solve the bandwidth constrained delay constrained path problem. Also, the simulations indicate that reasonable usage of traffic engineering can efficiently improve the performance of routing algorithms.更多还原