Data Link Control Layer Performance for Wireless ATM Networks

The growing demand for ATM-based technology and recent proliferation of wireless access technologies have motivated researchers to examine the feasibility of extending the ATM paradigm from the wireline to the wireless domain and create a new research area known as Wireless ATM (WATM) (Toh, 1997). Dealing with lossy wireless links, characterized by limited bandwidth and high, bursty error rates, breaks the main assumption of conventional ATM systems, which is that of using no errors per links. Therefore, WATM systems must provide a transparent mechanism to ensure reliable end-to-end data transmission over the wireless portion of the network. The identification of a wireless-specific data link control layer (W-DLC), sitting between the traditional ATM layer and a wireless-specific medium access control layer (WMAC), is the responsible entity for guaranteeing the quality of service (QOS) requested by individual ATM-based virtual connections. Thus the main focus is to investigate the performance of DLC protocol for ABR traffic over wireless ATM network. Retransmissions are only required for non-real time traffic and are implemented using a Go-Back-N and Selective Repeat (SR) ARQ (Lin et aI., 1984), (Schwartz, 1987). Wireless channels are usually time-varying and the channel bit error rates vary as the surrounding environment changes. Since these factors put in jeopardy the performance of the DLC protocol and higher layer end-to-end protocol at large, additional link-level mechanisms are added to provide reliability over impaired radio links. The DLC protocol implementation represents an attempt to achieve these goals under the strict constraints imposed by impaired wireless links. This thesis studies and compares the two Automatic Repeat Request (ARQ) protocols, i.e., Go-Back-N (GBN) and Selective Repeat (SR) ARQ and analyse them for variable packet size and fixed packet size (WATM packet) by using C programming for simulation. The results show that the performance of SR ARQ is better than the GBN ARQ for variable packet size. The results also show that SR ARQ protocol has better performance than GBN ARQ in terms of error detection for fixed WATM packet in the range of 50 -70 bytes, which is the WATM packet range (ATM Forum, 1997)