Dynamic Joint Network-Channel Coded Cooperation for Underwater Data Collection Yougan Chen, Xiaomei Xu, Haixin Sun, Lan Zhang, and Zhaotong Zhu Key Lab of Underwater Acoustic Communication and Marine Information Technology (Xiamen University), Ministry of Education, Xiamen 361005, PRC xmxu@xmu.edu.cn Abstract—In this paper, we propose dynamic joint network- channel coded cooperation (DJNCC), for an underwater data collection problem, where multiple underwater nodes need to send data to one common destination. DJNCC seamlessly couples channel coding and dynamic network coded cooperation (DNC) to effectively combat the detrimental effect of fading of underwater channels, utilizing the fact that underwater nodes can overhear the transmission of others and relay each other when necessary. The relay-nodes participating in the cooperation are selected by the destination based on the measured pilot signal to noise ratio of the underlying OFDM modulation. Through both analysis and simulation for one specific underwater topology, we demonstrate the significant performance improvement of DJNCC over other schemes. Keywords- Joint network-channle coded cooperation; OFDM; Dynamic coded cooperation;Underwater acoustic networks I. INTRODUCTION In large-scale underwater acoustic (UWA) sensor networks, to monitor phenomena in underwater environments, one research issue is to design effective approaches for one central node to collect the data from multiple underwater sensors [1-3]. Compared to wireless radio communications, UWA communications suffer from high and space-time-frequency- varying packet losses due to the detrimental effect of fading of UWA channels [4-6]. Some effective methods to provide reliable communication are using error correction codes inside a packet at physical layer (channel coding) or erasure correction codes across multiple packets at network layer (network coding) or both (joint network-channel coding) [7-8]. Existing topology networks related to the data collection design protocols can be roughly classified into the following two categories: homogeneous networks and heterogeneous networks. In the former case, all the communication channels, either from a terminal to the destination or between two terminals, are spatially independent and have the same signal- to-noise ratio (SNR). The adaptive network coded cooperation (ANCC) [1] and the generalized adaptive network coded cooperation (GANCC) [2] are proposed for this network where the coding structure matches well with the network topology. In the later case, selective relay cooperation (SRC) and dynamic network coded cooperation (DNC) [3] are proposed to combat the case when only parts of the nodes are dynamically participating in the cooperation phase instead of all due to individual channel characteristics (SNRs). However, if there was only one undecoded node left, the DNC scheme then deteriorates into SRC scheme and no more benefit can be obtained from joint iterative decoding among nodes, i.e., there is no network coding gain in this case. Actually, the higher collection rounds, the fewer undecoded nodes left, which lead to less network coding gain for DNC scheme. Hence we consider coupling the physical layer channel coding and the network layer DNC scheme to combat this issue. In this paper, we propose dynamic joint network-channel coded cooperation (DJNCC), to increase the overall system performances, combining the physical layer channel coding gain at the first round and the DNC scheme. At the physical layer, we also consider OFDM modulation [9] as our previous work in [3], where the SNR can be effectively measured at the pilot sub-carriers. The transmission schedules are also optimized by the destination based on the pilot SNRs from the nodes to the destination and the side information on what packets from other nodes are available at each relay. The main difference lies in the joint network-channel decoding at the destination. The rest of the paper is organized as follows. In Section II, we describe some preliminaries for the proposed DJNCC scheme, including the system model and the related SRC and DNC schemes. We then present the proposed scheme in Section III and show the simulation results of all the schemes as a comparison in Section IV. Finally the conclusions are drawn in Section V. II. PRELIMINARIES In this section, we first present the underwater data collection system model, and then present the background on the SRC and DNC schemes about data collection protocols that are previous work in our proposed DJNCC scheme. A. System Model The system includes N u terminals that communicate wirelessly to a common destination. Fig. 1 shows one example where 5 underwater nodes S 1 , S 2 , S 3 , S 4 , S 5 send data to a common surface destination D. For UWA transmissions, we assume that each node uses This work was supported by the National Natural Science Foundation of China Grant No. 41176032, the Fundamental Research Funds for the Central Universities Grant No. 201112G020 & No. 201212G012, and China Scholarship Council. 978-0-933957-40-4 ©2013 MTS