Date of Completion
Underwater Acoustic Networks; Reliable Data Transfer, Hybrid ARQ, Medium Access Control
Reda A. Ammar
Field of Study
Computer Science and Engineering
Doctor of Philosophy
As an emerging research area, Underwater Acoustic Networks (UANs) have attracted tremendous interests in last several years. Reliable and efficient data transfer is of critical importance for UANs since it provides essential services for various tasks. Due to the unique features of UANs including the long propagation delay, low bandwidth and high error probability, reliable and efficient data transfer has been facing great challenges. In this dissertation work, we tackle this problem from three different perspectives.
Hybrid ARQ (HARQ), which combines Forward Error Correction (FEC) Coding and Automatic Repeat reQuest (ARQ) has been proved to be an effective way to counter packet loss caused by error prone UAN channels. Along this path, we propose three HARQ based schemes. We focus on string topology unicast UANs, which have been widely adopted for underwater applications. What these three schemes have in common is that they all employ a combination of FEC coding and ARQ. However, they choose different FEC coding schemes based on the underlying UAN platforms and targeted application scenarios. Also different multi-hop coordination schemes are adopted to handle collisions.
Underwater Hybrid ARQ (UW-HARQ) is a simple end-to-end approach targeted for UANs with limited computation capability and power capacity. Random Binary Linear Coding is employed by UW-HARQ due to its low complexity and an optimal coding ratio estimation scheme is proposed. For powerful UAN platforms with more computation capacity and less energy constraint, we design a Coding based multi-hop Coordinated Reliable Data Transfer (CCRDT) protocol, which is tailored for multi-hop homogeneous string topology UANs. GF(256) Random Linear Coding is employed to enhance reliability and efficiency over one hop. A multi-hop coordination scheme is proposed to eliminate collisions and improve throughput. CCRDT has been implemented on real UAN nodes and extensive lab tests have been conducted to show its advantages over existing approaches. To work with heterogeneous UANs where channel quality on different links sometimes varies significantly, Fountain code based Adaptive multi-hop Reliable data transfer (FOCAR) is proposed. A multi-hop optimization approach is incorporated into FOCAR to achieve minimum end-to-end delay over multiple hops.
The three HARQ based approaches are targeted for string topology unicast UANs. To address the reliable and efficient broadcast problem for UANs, we propose a Two-phase Broadcast (TBS) scheme. TBS does not rely on topology or neighbor information and thus is more adaptive to the dynamic changes in UANs. TBS includes two phases: the Fast Spreading phase and the Data Recovery phase. The Fast Spreading phase combines opportunistic overhearing and network coding to improve broadcast efficiency. The Data Recovery phase aims to guarantee reliability and reduce interference probability.
CCRDT and FOCAR utilize multi-hop coordination to eliminate collisions in string topology UANs. By contrast, we propose a Selective ARQ and Slotted Handshake based Access (SASHA), for arbitrary topology UANs, which allows the occurrence of collisions but is able to recover from them. SASHA embraces the most commonly employed techniques in coordination based UAN MAC design. We implemented SASHA on real UAN nodes and conducted a sea test to evaluate its performance. We investigated the hop-by-hop and end-to-end behavior of SASHA. From the experimental data, practical issues have been discovered and corresponding design guidelines are suggested.
Mo, Haining, "Reliable and Efficient Data Transfer for Underwater Acoustic Networks" (2014). Doctoral Dissertations. 518.