Improvement And Optimization Of Mac Protocol For Large Scale Wireless Sensor Networks

In recent years, the broad application prospect of wireless sensor networks (WSNs) have raised a hot wave of research of domestic and foreign researchers. Large network size is inherent characteristics of WSNs. Many factors, like the increasing of node distribution density, number of nodes, and network traffic with the enlarging network size, aggravate the channel access confliction, and make the network performance worse dramatically. Around this serious problem, this paper focuses on the enhancement of MAC protocol and intends to establish a performance-improved and stable large-scale wireless sensor networks.Firstly, this paper proposes a new backoff algorithm for IEEE 802.15.4 MAC protocol, which is based on two-level channel information mined by nodes locally, namely channel busyness rate and packet collision rate. These two parameters are adjusted according to the Exponentially Weighted Moving Average method, based on which nodes adjust the backoff window properly to maximize the channel utilization. In order to avoid packet congestion, the backoff window of each node shrinks linearly as traffic load is sensed light, which makes transmission probability of nodes fluctuate around the optimal probability for p-persistent CSMA model. Compared with IEEE 802.15.4 standard and KBE algorithm, the simulation results on NS-2 platform show that our proposed scheme significantly improves the network performance, such as throughput and MAC delay, especially in the high-density and heavy load network.Secondly, this paper establishes a p-persistent CSMA evolutionary game model (EGP-CSMA) for IEEE 802.15.4 CSMA/CA protocol in the scenario where the wireless channel error may occur. In this model, the unique evolutionary stable strategy (ESS) is derived in the case of fixed number of nodes, and the optimal ESS to maximize saturation throughput and minimize average energy consumption are calculated respectively. Numerical analysis shows that, as the node number and bit error rate are fixed, the throughput can be maximized and the energy consumption can be minimized, as long as nodes adjust the transmission probability at optimalESS. When the network size is large, the throughput and energy consumption can be jointly optimized with a single cost rate. Besides, the model show excellent scalability of network size.Lastly, according to Lyapunov stability theory, this paper studies the stability at ESS of replicator dynamics with and without delay respectively. In this part, the asymptotic stability at ESS of replicator dynamics without delay are verified and sufficient conditions of stability at ESS of the delayed one are derived. Especially, we study the effect of payoff delay, success payoff and bit error rate on the convergence of evolutionary dynamics to the optimal ESS. Numerical simulation results in MATLAB show that, as the bit error rate is constant and payoff delay is short, a proper cost and payoff can make the multiple access game reach an evolutionary stable status at the optimal transmission probability. Thus, the p-persistent CSMA network can be stable and perform optimally.