Successive Slepian Subspace Projection in Time and Frequency for Time-Variant Channel Estimation

This paper describes an iterative time-variant channel estimator for the uplink of a multi-carrier code division multiple access (MC-CDMA) system. MC-CDMA is based on orthogonal frequency division multiplexing (OFDM). Due to OFDM every time-variant frequency-flat subcarrier can be estimated individually. We develop a successive Slepian subspace projection in the time and frequency domain. The subspace in the time domain is spanned by discrete prolate spheroidal (DPS) sequences. The DPS sequences are designed according to two system parameters, the maximum Doppler bandwidth and the block length. The correlation between the individual subcarriers can be exploited by a subspace projection in the frequency domain, using so called shifted DPS sequences. The shifted DPS sequences are designed according to the maximum essential support of the channel impulse response and the number of subcarriers. Due to the subspace projection in the frequency domain the presented channel estimator is able to deal with realistic channels that have non sample-spaced path delays. We validate the channel estimator with a geometry based stochastic channel model according to the COST 259 recommendations.