| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 93 === Due to amazing technology progress of wireless networks facilities in recent years, wireless networks become popular and popular in the world. Many products, such as notebooks and personal digital assistants, have built-in wireless networking devices. This makes the world raise a new trend. Many places have base stations for wireless accesses. Researches in the field of wireless networks become more and more important.
Though wireless networks provide much convenience, in some aspects they are worse than wired networks, such as two of the most important drawbacks -- limited bandwidth and limited energy. These drawbacks limit applications of wireless networks. Therefore, many people make some improvements in increasing throughputs and energy utilizations. However, most of these improvements are studied assuming data are transferred over the perfect network channels -- no noises in the network channels. However, in real world, wireless networks usually have serious packet loss problem due to high noises and many uncertain and un predictable factors such as the barrier of the buildings. Due to the packet losses in wireless networks, we must adapt the transmitted packets, such as the packets lengths. Therefore, in the thesis, we adapt the packet lengths according to different network conditions.
In this thesis, we consider adjusting packet length in wireless networks. When networks transmit packets, throughput and the dissipated energy are related to packets lengths. When we consider bit error rate (BER) and packet error rate (PER), we adapt the packet lengths according to error statistics and network conditions so that throughput and energy utilization are increased. Therefore, this saves great part of network bandwidth and energy wastes from packet losses.
In this thesis, we use the clustering-based wireless sensor networks (WSNs) as an application of the proposed method. The reason using WSNs as exemplar is that the WSNs usually have complicated network topology comprising a great deal of transmission types. Transmitting types include communications among sensor nodes, from sensor nodes to cluster heads (gateways), among gateways, from gateways to the remote base station, from sensor nodes directly to the base station, and so on. In addition, to present the applicability of the proposed method in mobile wireless networks, we also develop the equivalent distance model of noise error rates. We calculate the equivalent distance according to the network condition. In addition, through the equivalent distance model we reorganize clusters in the wireless sensor network and thus obtain a new topology. Then we perform energy proportional routing (EPR) for each newly formed topology every round. Since the equivalent distances among sensor nodes are time-variant, this implies that the sensor nodes are mobile. In contrast to conventional approaches that seldom discuss adaptation of packet lengths and variation of distances in wireless sensor networks, the proposed mechanism breaks through the restriction that the positions of sensor nodes are fixed.
In this thesis, we derive the optimal packet length formulae from some mathematical equations. Then we use NS2 to simulate the proposed model and list the simulation results. We verify the proposed model by comparing simulation results with mathematics analyses. The simulation results include throughput, energy utilization, and the total data amount received by the base station. From these simulations, we compare with the latest research studies related with wireless sensor networks as to prove that the proposed method is superior and beneficial.
|
|---|
