Joint channel estimation and data detection for OFDM systems over doubly selective channels

Abstract

In this paper, a joint channel estimation and data detection algorithm is proposed for OFDM systems under doubly selective channels (DSCs). After representing the DSC using Karhunen-Loève basis expansion model (K-L BEM), the proposed algorithm is developed based on the expectationmaximization (EM) algorithm. Basically, it is an iterative algorithm including two steps at each iteration. In the first step, the unknown coefficients in K-L BEM are first integrated out to obtain a function which only depends on data, and meanwhile, a maximum a posteriori (MAP) channel estimator is obtained. In the second step, data are directly detected by a novel approach based on the function obtained in the first step. Moreover, a Bayesian Cramer-Rao Lower Bound (BCRB) which is valid for any channel estimator is also derived to evaluate the performance of the proposed channel estimator. The effectiveness of the proposed algorithm is finally corroborated by simulation results. I. INTRODUCTIONWith high spectral efficiency and robustness against frequency selective fading, orthogonal frequency division multiplexing (OFDM) has been widely recognized as an efficient transmission technique for wireless communications and has been considered as a strong candidate for next generation mobile systems. Generally, in mobile systems high speed movement would cause Doppler spread and result in a multi-path time-varying channel, i.e., doubly selective channel (DSC). Due to the time variation of the channel, the number of channel parameters in one OFDM symbol significantly increases, which makes the channel estimation a challenge. Meanwhile, this doubly-selectivity in channel will destroy the orthogonality among subcarriers and induces intercarrier interference (ICI) in OFDM systems, which also complicates the data detection.So far, channel estimation and data detection for OFDM systems over DSCs are usually tackled separately. In the literature [1]-[6], a number of pilot-aided channel estimation methods have been proposed by adopting various kinds of basis expansion models (BEMs). With the channel estimate, a minimum mean square error (MMSE) data detection method and an improved version with low-complexity have been proposed in [5] [6] respectively. In recent years, joint channel estimation and data detection methods have been developed for single carrier systems in [7] [8]. It is shown that those joint schemes can generally provide better performance than the separate approaches, since the data are also exploited for channel estimation. However, their extension to OFDM systems is by no means straightforward.In this paper, a joint channel estimation and data detection algorithm is proposed for OFDM systems over DSCs. K-L

Joint Channel Estimation and Data Detection for Multihop OFDM Relaying System under Unknown Channel Orders and Doppler Frequencies

“…In the figures, KLEM represents the performance of the EM algorithm with channel expanded on Karhuen-Loève (KL) bases. This algorithm is an extension of the KLEM algorithm for singlehop case [23], and the detail is not included in this paper due to space limitation. The KLEM algorithm requires full information on channel tap positions, Doppler frequencies and power profile of each channel, together with noise statistics, thus serves as a reference for optimal performance here.…”

“…The KLEM algorithm requires full information on channel tap positions, Doppler frequencies and power profile of each channel, together with noise statistics, thus serves as a reference for optimal performance here. And CRLB curve represents the Cramé-Rao lower bound, which can be obtained from [23] by replacing the single-hop channel and noise power with the composite channel and composite noise power. Meanwhile, ideal case with full channel information at the receiver is also depicted as the performance bound in the BER figure.…”

“…and the form of Q functions are needed, in addition to (20), ( 21), ( 22) and (23). Since the noise is assumed to be AWGN, the likelihood function is given by…”

Joint Channel and Doppler Spread Estimation over Time-Varying Flat-Fading Channels

“…Techniques combining channel estimation with other channel-related processes, e.g., data detection [9], multi-user detection [10], carrier-frequency-offset (CFO) estimation [11] etc, have attracted many research attentions. There are few researches investigate joint estimation of channel and Doppler spread [12], [13].…”