Study on Connection Admission Control Method Based on Wireless Sensor Network

Abstract

In order to solve the overflow of aggregation node in wireless sensor network, a novel connection admission control method (Connection Admission Control based on Wavelet Transform, CACWT) is proposed. Based on the burst characteristic of wavelet transform reducing network service flow, border of effective bandwidth and overflow probability are derivated. Meanwhile, a simulation test with NS2 was conducted to probe the relationship between effective bandwidth and buffer zone. Compared with other traditional algorithms, CACWT has better suitability. Introduction Wireless sensor network, one of the supporting technologies of the Internet of things, is constructed by a wide range of low-cost micro sensor nodes, collaboratively accomplishing the task of collecting, transmission and disposing of perceptual information in the deployment area. It has many characteristics that differ from the traditional wireless network, and can be widely used in many fields such as military affairs, target tracking, environmental monitoring, medical aid, space exploration, precision farming, industrial automation and so on, which has received widespread attention from the industry and academia. With the fast popularization of wireless sensor network, how to gurantee the performance of service flows has increasingly become a key issue of concern. Admission control is the key technology to ensure the optimizaion of the performace of service flows, it is formed by three parts: service flow description, admission criteria and measuring process. A typical example of the service flow description mechanism is token bucket, and admission criteria is mainly of bandwidth constraint, equivalent capacity and sum-rate, while measuring process usually adopts time window, point sampling and exponential averaging. In the process of carrying out the connection Admission Control, the network, based on the features of network resources and service flows, decides whether to accept the connection request of the user. Considering such a point, academics both at home and abroad have proposed quite a few evaluating algorithms and evaluation criterions, such as connection admission control based on parameter, connection admission control of end to end statistical model, the dynamic bandwidth reserved mechanism and flow admission control based on distributed measurement, adaptive measurement-based admission control and so on. A novel connection admission control method is put forward based on the fractal character of service flow. In this method, the burst characteristic of service flow is reduced by wavelet transform at first, and utilizes Fractional Brownian Motion provides that aggregation node of the border of International Conference on Computational Science and Engineering (ICCSE 2015) © 2015. The authors Published by Atlantis Press 517 effective bandwidth and overflow probability, then a simulation was conduct to deeply analysis the performance conditions of connection admission control. Introduction of model All the nodes in a wireless sensor network (its topological structure shown in Figure 1) are arranged in the same wireless Ad hoc network. Supporting node K as aggregation node, the rest nodes N will be data source nodes or relay nodes. In order to copy with the phenomenon of network congestion in time, the performance of network at node K need to be evaluated. Meanwhile, recently study shows, actual service flows have fractal character and self-similar character. Thus, it needs to combine with this character to build connection admission control method of service flows of node K. The representative method of describing the fractal character of service flow is Fractional Brownian Motion(Fractional Brownian Motion,FBM), defined as: H A(t) mt amB (t)   (1) While, A(t) is the aggregating service flows within the moment of time t ,  shows as a coefficient of variation, m shows as average arrival rate, H shows as self-similarity(0<H<1), BH(t) is Norm Fractional Brownian Motion. Supposing system buffer zone b approach to infinite, if the service rate at node k, i.e. bandwidth λ is greater than the average arrival rate m of service flows, the length Q at node k needs to satisfy: s t Q(t) sup(A(t) A(s) (t s))       (2) The distribution of the length can be marked as: H 1 H ( m) q P{Q(t) q} k(H) am             (3) And, ()  is residual function, H H H H H k    1 ) 1 ( ) ( Supposing the overflow probability at node K is η, according to the proposed large deviation theory of document[18], the length of infinite buffer greater than the formula of probability calculation b shall satisfy: