Low Complexity Finite-Alphabet based Blind Channel Estimation in OFDM Receivers

In this paper, we present two novel blind channel estimation algorithms for OFDM systems based upon the knowledge that the modulated and transmitted data are confined to a finite alphabet set. The Minimum Impulse Length (MIL) method exploits the correlations between adjacent subcarrier coefficients caused by time limited impulse responses and shows a very good estimation performance. Unfortunately, MIL and most other FiniteAlphabet based blind channel estimation approaches are known to be extremely complex due to an exhaustive search to be performed over a tremendous number of channel coefficient combinations. Our Clustered SubCarriers (CSC) method, however, dramatically reduces this number of coefficient combinations to be checked without a significant deterioration in estimation quality. Using bit error rates (BERs), both algorithm are tested with simulations and compared to other blind and nonblind channel estimators. Keywords— OFDM, Blind Channel Estimation, Finite Alphabet. I.INTRODUCTION In recent years, Orthogonal Frequency Division Multiplexing (OFDM) has become one of the most important techniques for high-rate wireless data transmission. Especially, it has been proposed for the European HIPERLAN/2 standard and the American equivalent IEEE802.11a, which are two similar concepts for broadband wireless local area networks (WLAN) in the 5 GHz band. Since these standards include coherent data demodulation, the transmission channel has to be estimated. Generally, this is achieved by nonblind channel estimators exploiting additionally transmitted training data. Moreover, the training sequences have to be transmitted periodically, since the channel in wireless applications normally is time variant. In order to increase bandwidth efficiency, blind channel estimation, on the other hand, is well motivated since it avoids the need of any training data. The so-called Minimum Distance (MD) This research was supported by the German NSF (DFG contr. #Ka 841/5-1). algorithm [1] is a blind estimator which is based on the knowledge that the modulated and transmitted data are confined to a finite alphabet set. In this paper, we present the new Minimum Impulse Length (MIL) approach, which shows a slightly better estimation performance than MD. However, both MIL and MD require an enormous computational effort. In contrast, our Clustered SubCarriers (CSC) scheme dramatically reduces this effort without any significant deterioration in estimation quality. The paper is organized as follows: Section II gives an overview of the OFDM system. In section III, we present our new blind channel estimation algorithms. After showing some simulation results in section IV, the paper is concluded in section V. II.DATA TRANSMISSION IN OFDM SYSTEMS Figure 1 shows the conventional OFDM system with Cyclic Prefix (CP). The CP of length Ncp larger than the channel order q avoids the received data from being disturbed by inter-symbol (ISI) or intercarrier interference (ICI). In the transmitter, the chanChan. encod. S/P Mod. Mod. IFFT & Cyclic Prefix u k ( d i ( ) 0 d i ( ) N-1 Mod. d i ( ) 1 OFDM Transmitter