Final report of the coding/decoding algorithms and performance comparison : ICT-619555 RESCUE D2.1.2 Version 1.0

Abstract

<div lang="en" class="abs"><p class="abs_header">Abstract</p><p>This deliverable provides comprehensive analytical and simulation results of the coding/decoding algorithms used in RESCUE. The algorithms are applied to various frameworks and investigated the applicability in selected cases. We also address the impact of lossy forwarding (LF) relaying in fading channels and specifically consider the evaluation of frame error rates (FERs) in comparison with lossless decode-andforward (DF). Moreover, the deliverable proposes an algorithm for opportunistic relay selection in LF-based networks where the best relay can be found using a machine learning method. Remarkable performance gains are observed in a case where LF concept is applied in multi-antenna scenario, in which relays can transmit and receive at the same time by using multiple antennas for interference cancelation. Furthermore, we apply joint decoding (JD) into IEEE 802.11WiFi standard. In addition, we present an error rate model for RESCUE physical layer to be used when designing higher layer protocols. After analyzing the results presented in this deliverable, we conclude that RESCUE concept is especially suitable for the cases where the retransmissions are not preferable or not even possible.</p></div>

<div lang="en" class="abs"><p class="abs_header">Executive summary</p><p>The RESCUE project — “Links-on -the-fly Technology for Robust, Efficient, and Smart Communication in Unpredictable environments” — proposes a novel multi-route communication technology for multi-hop networks that are subject to dynamic topology changes. RESCUE is a cooperative communication paradigm consisting of decodeand-forward (DF) relaying that allows intra-link errors, combined with distributed turbo coding (DTC) that brings improved error protection. The relaying nodes decode and re-encode the data, and even with possible decoding errors, there exist correlation between the forwarded packet copies because they originate from the same source. Therefore, destination can improve the reliability of the detection by taking the correlation into account when decoding. The relaying method considered in RESCUE is referred as lossy forwarding (LF) which, in its original form, means that no integrity check is performed at the relay, but the frame is always forwarded. In this deliverable, we provide comprehensive analytical and simulation results of the coding algorithms used in RESCUE. Furthermore, we apply the coding algorithms to various frameworks and investigate its applicability in selected cases. The gains of the forward error correction (FEC) coding and joint decoding (JD) used in RESCUE are evident already from D2.1.1. Therefore, in this deliverable we also address the impact of LF relaying by comparing frame error rate (FER) results with the lossless counterpart that uses the same coding/decoding.</p><p>Firstly, we investigate a three-node system, referred as toy scenario 1 (TS1) defined in D1.2.1, by briefly introducing theoretical results considering the outage probabilities of different relaying strategies including a feedbackbased strategy. Furthermore, we deeply investigate the differences between theoretical and practical simulation results, aiming at finding possible directions for improving the coding/decoding scheme. The second investigated case is a system with multiple relays without direct link, referred as toy scenario 2 (TS2), for which we show the superiority of LF compared with lossless forwarding in specific cases.</p><p>Opportunistic relaying is an attractive solution in order to reduce system complexity. In opportunistic relaying, the best relay is selected among available relay candidates according to a certain policy. It has been shown that there is no loss in performance in terms of the diversity-multiplexing tradeoff if only the best relay participates in cooperation. In this deliverable, we present an opportunistic relay selection strategy for LF system. Furthermore, a statistical learning method is proposed for practical implementations.</p><p>A successive single-input multiple-output (SIMO) erroneous relaying with iterative maximum a posteriori (MAP) receiver at destination is also proposed in order to allow relays’ transmission and reception at the same time. In other words, one of the relays can transmit and the other receive resulting in the reduced multiplexing loss. The proposed scheme allows relays erroneously forward the decoded symbols, and jointly cancels interference and combines diversity at destination. It has been shown that, without extra cost of time delay and signalling overhead, the proposed scheme can exhibit up to 8 dB gain by comparing with the conventional selective DF (S-DF) based schemes.</p><p>An intuitive application of RESCUE contributions into current wireless standards is the use of JD in hybrid automatic repeat request (HARQ) to combine the retransmitted erroneous packets at the destination. In this deliverable, we report the results of applying JD into IEEE 802.11 WiFi standard. We show that with a minor changes, we can achieve performance improvements in terms of FER.</p><p>In order to efficiently evaluate and test the higher layer protocols, the abstraction in the physical layer (PHY) is needed. Therefore, we present an error rate model for RESCUE PHY, which is capable to accurately predict the link-level performance at low computational complexity and can therefore be utilized for the design and performance evaluation of algorithms and protocols that exploit DTC as well as for system-level simulation.</p><p>At the end, we find that LF is beneficial at least in the following cases:</p><p>1. Communications in rapidly changing network topology where the retransmissions are not possible.</p><p>2. Communications with low power devices where the nodes need to share the load with each other.</p><p>For FEC coding/decoding used in RESCUE, we can conclude that it is very simple but remarkably strong performing close to the theoretical limits. Therefore, it has a great potential to be applied in current wireless standards.</p></div>