Modeling Hybrid Access Mechanism Of IEEE 802.15.7 And Performance Evaluation

With the rapid increase of the demand on communication rate, RF communication can not meet people’s needs for bandwidths. Visible light communication(VLC) attracts worldwide attention and has become one of the most important next-generation wireless communication network as its advantages of high bandwidth and unregulated spectrum. In recent years, VLC have attracted much attention. Most of the researches concentrate on the techniques of physical layer and the research on MAC layer is rare. Traffic flows have heterogeneous quality of service(Qo S). One of the key problems of the future communication network is how to support traffic flows with Qo S guarantees efficiently.IEEE802.15.7 is the standard for short-range optical wireless communications using visible light. It provides hybrid access mechanism for heterogeneous services. CSMA/CA mechanism is used in contention access period(CAP) while guaranteed time slots(GTS) are allocated in contention free period(CFP). The hybrid access mechanism can exactly offer Qo S guarantee for heterogeneous services. In wireless communication, the random service mechanism and the time-variant characteristic of wireless channel can both cause random service, which makes it different to guarantee the deterministic Qo S requirements. Therefore,statistical Qo S guarantee attracts a wide spread attention. It is significative to research on guaranteeing statistical Qo S in VPAN. Effective capacity is a link-layer channel model that can support statistical Qo S requirements as it jointly consider the physical layer and the Qo S metrics of link layer. Therefore, we apply the effective capacity theory to support the statistical Qo S requirement. In this paper, we model and analyze the hybrid access process of IEEE802.15.7.Then we analyze the effective capacity for heterogeneous traffic under the service process of IEEE802.15.7. The results provide a theoretical reference for statistical Qo S guarantee in VLC network.Under the background of single non-real-time service, terminal only accesses the network in CAP. The limitation of CAP length will cause the deferring of random backoff process. Considering that constraint, the random backoff process for non-real-time service is modeled as a three-dimensional Markov chain. We analyze performance indexes in terms of packet drop probability and throughput. Then, a transfer function diagram of random backoff process in z domain is investigated to solve the distribution of random backoff sojourn time.We establish a semi-Markov on-off model to model the service process of non-real-time service. The effective capacity is solved by analyzing the moment generating function of on period and off period respectively.Lastly, we analyze the influences on effective capacity made by network metrics and solve the buffer overflow probability under different traffic flow distribution by simulation.Under the background of heterogeneous services, non-real-time services only participate in the CAP while real-time services participate in both CAP and CFP. Firstly, a three-dimensional Markov chain is established to investigate the random backoff process for heterogeneous services. We analyze the average number of successfully received packets in CAP. Secondly, we model the GTS allocation process as an one-dimensional Markov chain.The PDF of successfully received GTS requests is used as the link to establish the coupling relationship between the two period. The dynamic change process of CAP and CFP is analyzed, and the convergence length of CAP is derived. Besides, we investigate a transfer function diagram of random backoff process in z domain to solve the distribution of random backoff sojourn time.Then we study the steady state distribution of GTS request number in buffer under the condition of convergence length and derive the waiting time distribution of GTS requests. The service process of non-real-time services and real-time services are modeled as two different semi-Markov on-off model and effective capacity are solved respectively. We also analyze the overflow probability of different services. Lastly, we research how different network metrics influence effective capacity and establish a sum effective capacity function for a terminal by simulation. Effective methods to improve the effective capacity are provided to ensure the statistical Qo S.