Effects of Imperfect Subcarrier SNR Information on Adaptive Bit Loading Algorithms for Multicarrier Systems Alexander M. Wyglinski Fabrice Labeau Peter Kabal Electrical & Computer Engineering, McGill University 3480 University St., Montr´ eal, Qu´ ebec, Canada H3A 2A7 Email: alexw@TSP.ECE.McGill.CA Abstract – In this paper, we evaluate and compare the robust- ness of several adaptive bit loading algorithms for multicarrier transmission systems when imperfect subcarrier signal-to-noise ratio (SNR) information is used. In particular, we investigate the impact of the uncertainty of data-aided channel estimation tech- niques on system performance. We also examine an implemen- tation issue associated with adaptive bit loading algorithms that use metrics related to the SNR. Although such metrics can be de- rived via closed form expressions, look-up tables are used instead to reduce system complexity, resulting in the SNR values being quantized. Thus, we examine the effects of SNR quantization on system performance. Finally, we present a technique for choosing SNR values in a fixed length look-up table in order to minimize quantization error. Keywords: Adaptive modulation, multicarrier communications, imperfect channel estimation I. I NTRODUCTION Multicarrier modulation (MCM) is increasingly being em- ployed in many high-speed data transmission systems, includ- ing several wireless local area network (WLAN) standards [1]. The advantage of MCM is that it can transmit data at lower rates on each subcarrier simultaneously. As a result, the frequency-selective fading channel is effectively transformed into a collection of nearly flat-fading subchannels. Many of these MCM systems, especially the wireless ones [1], use con- ventional multicarrier modulation which employs the same sig- nal constellation across all subcarriers. These suffer from the subcarriers with the poorest error performance. One solution is to perform adaptive “bit loading”, where the signal constel- lation size across the subcarriers varies. In extreme situations, some subcarriers can be “turned off” or nulled if the subcarrier SNR values are poor. There have been numerous studies on the performance of multicarrier systems that employ adaptive bit loading algo- rithms, where the subcarrier SNR information is assumed to be perfectly known [2–6]. However, these results are overly optimistic since they neglected the degree of uncertainty that exists with the subcarrier SNR information. In an attempt to provide more accurate results, several studies have included models of these sources of uncertainty, such as noisy chan- nel estimates [7, 8] and outdated channel estimates due to time varying channels [8, 9]. This research was partially funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Le Fonds de Recherche sur la Na- ture et les Technologies du Qu´ ebec. N N N Loading Algorithm + Adaptive Bit Demultiplexer () 0 () x n Σ channel state information feedback subcarrier bit allocation (1) ( ) x n () 0 () y n (1) ( ) y n () 0 () g n (1) ( ) g n ( 1) () N g n - () hn ( ) xn () rn () vn (a) Transmitter with adaptive bit loading algorithm and channel Channel Estimator Adaptive Bit Multiplexer Freq. Domain Equalizer N N N () r n () 0 () f n (1) () f n ( 1) () N f n - () 0 () y n (1) () y n () 0 () x n (1) () x n () x n ^ ^ ^ ^ ^ (b) Receiver with channel estimator Fig. 1 Schematic of an adaptive multicarrier modulation system in the downlink direction using feedback from the channel estimator. In this paper, we present a comparative study of the robust- ness of four adaptive bit loading algorithms with imperfect subcarrier SNR information. The adaptive MCM system used in this work is presented in Section II while the channel esti- mation technique and channel estimation error model are pre- sented in Sections III and IV. Section VI briefly describes the adaptive bit loading algorithms studied in this paper. Since adaptive bit loading algorithms use metrics that are functions of the subcarrier SNR, such the bit error rate (BER) [4–6], a look-up table of metric values can be used to reduce complexity. However, the size of the look-up table as well as the metric values chosen to be stored can impact system performance by introducing quantization error into the subcar- rier SNR information. In this work, the effects of subcarrier SNR quantization due to a finite-size look-up table is stud- ied. Section IV presents the SNR quantization noise model employed in this paper. Furthermore, a technique is presented in Section V that chooses the subcarrier SNR/BER pairs for the look-up table given a number of modulation schemes in or- der to minimize the quantization error. The simulation results Globecom 2004 3835 0-7803-8794-5/04/$20.00 © 2004 IEEE IEEE Communications Society Proc. IEEE Global Telecommun. Conf. (Dallas, TX), pp. 3835-3839, Nov. 2004