利用空間-時間關係性於水底感測器網路上之設計

Abstract

Since data in Underwater Sensor Networks (UWSNs) is transmitted by acoustic signals, the characteristics of a UWSN are different from those of a terrestrial sensor network. Specifically, due to the high propagation delay and the limited channel bandwidth of acoustic signals in UWSNs, current terrestrial approaches do not work well in UWSNs. Recently, there have been a variety of UWSN applications with different characteristics and goals, so this work focus on energy-efficient UWSN design for three different categories of UWSN applications: throughput-intensive, fair-data-collection, and delaysensitive applications. That is, we propose 3 different cross-layer UWSN modules: (a) an energy efficient MAC protocol for throughput-intensive applications, (b) a max-min fairness rate allocation scheme for fair-data-collection applications, and (c) an delay-aware opportunity-based routing approach for delay-sensitive applications. For the sake of energy-efficient design, our idea is to first develop Spatial-Temporal Conflict Graph (ST-CG), which describes the conflict relationship among transmission links explicitly by considering the Spatial-Temporal Relationship, and is able to avoid collisions to perform efficient energy consumption. Based on ST-CG, we consider Spatial-Temporal Relationship in the design of a bandwidth-efficient TDMA-based MAC protocol for UWSNs. In order to obtain a theoretical bound for the TDMA-based MAC schedules of ST-CG, a Mixed Integer Linear Programming (MILP) model is derived. To maximize the channel utilization for throughput-intensive applications, the TDMA-based scheduling problem in UWSNs is translated into a special vertex-coloring problem in the context of ST-CG. Then, we propose two novel heuristic approachs: (a) Traffic-based One-step Trial Approach (TOTA) to solve the coloring problem of ST-CG in a centralized fashion; and for scalability, (b) Distributed Traffic-based One-step Trial Approach (DTOTA) to assign data schedule in a distributed manner. Besides, a comprehensive performance study is presented, showing that the proposed MAC schedules TOTA and DTOTA can guarantee collision-free transmission and perform better than existing MAC schemes (such as S-MAC, ECDiG, and T-Lohi) in terms of the network throughput and energy consumption. For fair-data-collection applications, we study themax-min fairness problemin UWSNs. Time Expanded Clique (TiE-Clique) is proposed to represent the clique relationships with the Spatial-Temporal Relationship. We also devise an algorithm and integer linear programming model to assign max-min fair rates. The simulation results demonstrate that, the proposed time expanded solution is able to achieve max-min fairness and have higher system throughput. Finally, we exploit the idea of opportunistic-based routing to satisfy the requirements of delay-sensitive applications. Hence, by considering the propagation delays, a new routing scheme UWOR is proposed. Through extensive evaluations, we show that UWOR can improve the end-to-end goodput under deadline constraints.