Collision-Aware Approaches to Energy-Efficient Data Transmission in Sensor Networks

• Main Authors: Chun-HaoYang, 楊鈞豪
• Other Authors: Kuo-Feng Ssu
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/82477969262439005007

博士 === 國立成功大學 === 電腦與通信工程研究所 === 103 === The collisions of packet transmissions in sensor networks have been one of the main topics in MAC/Protocol layers. In wireless sensor networks, among all the solutions to the packet interferences, rescheduling has a potential to be a profitable solution which has abundant issues yet to be explored. The local rescheduling problem in wireless sensor networks has been firstly addressed and investigated in the thesis. The algorithms of local rescheduling have been proposed and evaluates the performance of reschedule solutions with different metrics. All solutions have to be under the limitation that the network should stay connected after the process of rescheduling. This thesis introduces a theoretical bound of maximum degree after node insertion. Along with empirical results in real world settings, the results motivate the design of algorithms and give possible reasons why existing rescheduling algorithms do not work efficiently. Two local link rescheduling algorithms and one local broadcast rescheduling algorithm are developed as improvements. Consider different cycle lengths, slot requests, and occupied slot number with different node densities and other critical parameters, simulations show that the developed algorithms greatly improve the ratio of finding proper solutions successfully in both types of scheduling compared with existing simple algorithms. Without breaking the network connectivity, 30% of successful rate is increased in link scheduling; 90% of successful ratio can be reached if the cycle length of the scheduling is long enough in broadcast scheduling. If the prescheduled network has sufficient occupied slots to release to the new node, nearly 100% of successful rate of rescheduling can be accomplished if the limitation is relaxed, where only network connectivity is guaranteed. Due to the unstable link connectivity and different characteristics in acoustic channel in underwater sensor networks, the scheme of local rescheduling is no longer beneficial to the design of the MAC/Protocol layers in underwater sensor networks. Given the properties of intermittent link connectivity in USNs, the existing message-based and synchronization-based approaches cannot meet packet delivery requirements. In recent years, a specific field of three-dimensional (3D) underwater sensor networks (USNs) have received substantial attention as a promising tool for target tracking and remote monitoring under the seas. Energy consumption is crucial in USNs since it is nearly impossible to recharge the batteries of the sensors. To deal with the collision issue and further reduce the energy consumption, an analysis for the probability of collisions between any two transmissions in USNs is presented in the thesis, in which the analyzed collision rate corresponds to the simulation and is demonstrated to be relaxed with a sufficient data processing rate in underwater networks. Under 10% of packet collision rate is found due to higher bandwidth and lower propagation speed in the water. Based on this result, a tailored delay-aware energy-efficient routing protocol (DEEP) is proposed for USNs. DEEP is composed of an energy model with realistic parameters in which the available 3dB bandwidth is derived with respect to the distances between nodes. DEEP involves an adaptable forwarding node selection mechanism, which incorporates the concept of energy efficiency and further reduces the collision rate. Simulations show that DEEP expends less energy for higher successful packet delivery compared with previous studies. Benefits from the higher bandwidth in USNs, DEEP reduces the collision occurrences and elevates 20% of the packet delivery ratio with the reduction of 30% of end-to-end delay time especially when the network conditions are unfavorable with average 20% of link quality. These results confirm that DEEP effectively handles the challenges.