Adaptive Puncturing and Rate Selection in Single-Codeword Turbo-Coded OFDMA Mai Abdelhakim Center for Wireless Studies Faculty of Engineering Cairo University Egypt Mohammed Nafie Wireless Intelligent Networks Center Nile University Egypt Ahmed Shalash Center for Wireless Studies Faculty of Engineering Cairo University Egypt Ayman Y. Elezabi Electronics Engineering. Department American University in Cairo Egypt Abstract This paper proposes using adaptive puncturing for rate-adaptive OFDMA systems utilizing turbo codes. The scheme is based on adaptively puncturing a Single Code Word (SCW) and hence adaptively changing the rate within the codeword. We compare the SCW against the Multiple Code-Words (MCWs) scheme where different rates are obtained by separate encoding, puncturing, and interleaving on a per-tile basis. Noticeable gains are obtained over the MCW scheme due to the use of larger turbo block sizes and hence larger interleavers. The SCW has around 1dB gain in goodput compared to MCWs, with much improved BER performance. We also propose a novel rate selection scheme for the SCW which we call Recursive Mutual Information Effective SNR mapping (R-MIESM) and compare it against the conventional MIESM scheme. The R-MIESM provides further goodput performance gains for the SCW compared to the MIESM method. I. Introduction Due to the varying nature of wireless channels, it is inefficient to have fixed transmission parameters during wireless communications. Allocation of resources should be adapted according to the channel variations in order to have efficient utilization of the available spectrum and consequently improve overall system throughput. Various schemes for adapting to the wireless environment have been studied in the literature. In [1], changing the modulation during transmission according to the channel quality is proposed, where 5dB gain over fixed transmission is obtained. In [2], Chua and Goldsmith proposed a variable rate variable transmit power adaptive scheme which obtains 5-10dB gain over the variable power fixed rate scheme, and 20dB gain over the non adaptive transmission. Adaptive modulation and coding (AMC) offers higher performance gains. Adaptive trellis-coded modulation and turbo-coded modulation were studied in [3] and [4] respectively, and it was shown that 3dB gain is obtained by the adaptive turbo-coded modulation over the adaptive eight-state trellis-coded modulation [4]. AMC is employed in OFDM systems where a sub- carrier by sub-carrier loading is proposed in [5] and the modulation order is varied for each sub-carrier according to the channel frequency response. The sub- carrier loading schemes improve the performance significantly compared to the non-adaptive OFDM systems. However, a huge overhead is required for such schemes. Less efficient, though still powerful, loading is to have a tile (resource blocks) adaptation, where a tile is a group of sub-carriers over consecutive OFDM symbols [6 – 8]. In adaptation on a per-tile basis, also referred to as zone loading, the transmission parameters are varied from one tile to the other according to the channel quality of each tile. In [6], adaptive modulation over tiles is introduced. In [7-8] adapting the code rate and the modulation scheme for each tile is investigated. In [7-8], the Multiple Codewords (MCWs) structure is used where the codeword size is confined to the tile size and several codewords are transmitted for each OFDM frame. Having the code rate varies on a per tile basis requires separate encoders for each tile. We proposed a novel architecture for changing the code rate in a rate-adaptive coded OFDMA system, via adaptive puncturing in [9]. In the proposed structure different transmission rates over tiles are obtained by using Single CodeWord (SCW) that spans the whole OFDM frame. The single codeword is adaptively punctured according to the required rates over the tiles. The SCW architecture resulted in significant performance improvement over MCW as well as Per- Frame Adaptation (PFA) when using convolutional codes [9]. The gains in that case were due to the increased frequency diversity afforded by the SCW architecture. In this paper we further investigate the performance of the SCW structure using Convolutional Turbo Codes (CTC). The SCW This work was supported in part by a grant from the Egyptian National Telecommunications Regulatory Authority (NTRA). 2009 IEEE Mobile WiMAX Symposium 978-0-7695-3719-1/09 $25.00 © 2009 IEEE DOI 10.1109/MWS.2009.31 174