Wireless data transmission: Error characterization and performance analysis

This dissertation addresses the accuracy of the data delivered from the transport layer and the connection performance in wireless channels. Our main purpose is to study these issues under network phenomenon/problems noticed recently and for wireless sensor networks which have become feasible only after the recent advances in IC fabrication technology, digital electronics, and wireless communications.; First, we analyze the error detection performance of 1's complement checksum. Internet checksum uses 1's complement arithmetic to detect errors in the content delivered by the data-link layer. Previous work on this topic determined the number of error passing patterns and the probability for only 2 and 3 bit errors, and the method used for determining the probability is hard to extend to more bit errors. We present a method to generate the formula of error passing probability. We then summarize some features of the probability. We also compare the performance of 1's complement checksum and 2's complement checksum.; Second, we address the problems with existing header compression algorithms and propose an adaptive header compression algorithm. Performing TCP/IP header compression on wireless links is necessary due large IPv6 header as well as the limited resources of wireless channels. However, people recently noticed high frequency packet reordering, and packet errors that avoid link layer error detection. We analyze the influence of these problems on existing header compression algorithms. We also propose a new algorithm, which is adaptive to the wireless channel as well as the packet size.; Third, we study the connection performance of wireless channels. Wireless channels usually face bursty errors, i.e., errors are prone to occur in clusters. These bit errors can be modeled as a discrete time Markov chain. Packet error statistics have also been modeled as a DTMC in previous work. However, whether this Markov chain is time-homogeneous has never been addressed. We prove that the packet errors can be modeled only as a Markov chain without constant transition probabilities, which means the Markov chain is not time-homogeneous. Thus finding a constant transition matrix and then discussing the performance is not accurate. Instead, some other models are proposed and the packet error/error-free length distributions are thus analyzed. We then use these models to analyze the ARQ performance for wireless channels. We also set up a gap model for FEC used in wireless sensor networks, where packet interleaving may not be adopted. The FEC performance is then analyzed based on this model.