LTE Uplink Channel Estimation Based On Compressed Sensing

The Long Term Evolution (LTE) is a project as the next generationcommunication systems launched by the Third Generation Partnership Project(3GPP) organization in order to satisfy the increasing requirements of multimediaservices. In China, the LTE technique based on time division multiplexing is themainstream development direction, which takes Single Carrier Frequency DivisionMultiple Access (SC-FDMA) as the transmission scheme. Channel estimation iscritical for the radio communication systems because of its role to support systemequalization and coherent demodulation. Traditional channel estimation algorithmsare researched in this paper. However, traditional algorithms based on fixed pilotstructure are confined to balance estimation accuracy and computation complexity,overlooking bandwidth efficiency. The channel estimation based on compresssensing proposed in this paper greatly reduces the pilot numbers under the premiseof good estimation performance. If LTE wants to guarantee the quality oftransmission and exerts its superiority, it must proceed accurate and efficientchannel estimation.First, this thesis put the Least Square (LS) estimation criterion intooptimization with adding zero in time-domain, and the Minimum Mean SquareError (MMSE) estimation criterion into simplification with Singular ValueDecomposition(SVD).Then, a series of methods, including estimation criterionsand interpolations to get channel information are simulated in order to provide areference for LTE channel scenarios. For the LS algorithm is the mainstream inLTE uplink and its poor estimation performance, the dual noise reduction LSalgorithm based the combination of transformation domain and threshold isproposed, and the simulation result demonstrates its superiority.As the latest research achievement in signal processing field, compresssensing theory implements to reconstruction sparse signal by little sampled valueswith high probability. At the same time, radio communication channel has thecharacteristic of sparsity. The traditional channel estimation algorithms do not takeadvantage of it. The thesis attempts to apply compress sensing theory to channelestimation for LTE-uplink. After proving its feasibility by theory, the effectivenessis confirmed by simulation. A channel estimation model based on compresssensing is presented, and an estimation algorithm with Orthogonal MatchingPursuit to get channel time domain impulse is proposed. Furthermore, the MSE,BER and throughput performances are obtained by simulation. Compared with thetraditional channel estimation algorithms, the CS channel estimation algorithms we proposed in this thesis have the advantages of less pilot numbers, betterbandwidth efficiency in the premise of the same precise estimation.