On the efficiency and performance evaluation of the bandwidth clustering scheme for adaptive and reliable resource allocation

A variety of disciplines have recently advocated the use of self-adaptive and auto-configuration methods, including biodynamics, cybernetics and computer modeling. Of these methods, one which exhibits numerous powerful features that are desirable in communication systems is adaptive swarm based intelligence. Swarm-based self-configuration does not require the need of external help, supervision or control. The stochastic nature of random events adversely affects the complexity of optimization tasks. This work proposes a bandwidth clustering scheme suited for this network resource allocation problem. Bandwidth clustering is used in a swarm-based active network environment where active packets continuously communicate with active nodes by using the Split Agent Routing Technique (SART). This mechanism enables the adaptation of the system to new conditions (bandwidth reservation/capacity allocation), as well as the passing of additional information to neighboring nodes in which the information is held in transmitted packets. Paths are clustered with respect to different levels of bandwidth in order to enable capacity allocation and bandwidth reservation on demand, for any requested traffic. The performance, reliability and adaptivity degree of the proposed scheme is thoroughly examined for different traffic measures, as well as the corresponding QoS offered (in terms of the end-to-end delay, available bandwidth and probability of packet loss). This scheme offers a decentralized and non-path-oriented way to efficiently increase the overall network utilization, enabling an equal share of network resources at the same time.