Research Of Channel Estimation Algorithm In MIMO-OFDM System Based On Compressed Sensing

Since the birth of human, we have never given up the study of communication technology. In December 2004, at the conference of 3GPP in Toronto, LTE(Long Term Evolution) officially started. It is the marking of the 4th generation mobile communication(4G) technology era has arrived.LTE is the evolution of 3G technology, which takes the OFDM(Orthogonal Frequency Division Multiplexing) and the MIMO(Multiple Input Multiple Output) technology as the only standard. It is currently the most advanced technology of civil mobile communication. MIMO-OFDM wireless communication system with high channel capacity and spectrum efficiency has been widely used in the occasions which need high speed and large amount of data. This system has obvious advantages, but the disadvantages that the acquirement of information is extremely dependent on the state of channel also can't be ignored. For the channel estimation is a critical step in MIMO-OFDM wireless communication system, whether the algorithm of it is good will directly affect the performance of the system.The so-called channel estimation is the use of pilots to samples wireless channel and applies reconstruction algorithms in the receiver to calculate channel parameters. The effect of channel estimation determines the quality of received signals. Unlike fixed wired channels, the wireless ones are so complicated that the parameters of them are not predictable. Under some circumstances, performance of the systems which are lack of channel estimation will be affected severely.The deployment of pilots must satisfy the Nyquist theorem in traditional channel estimation algorithms of MIMO-OFDM system. For rapidly changed radio channel, this means there will be a large amount of pilots need to be transferred and costing much frequency resources. Although the OFDM technology can improve the system spectrum efficiency, these saved spectra would be wasted for pilots do not carry any information. This runs counter to the purpose of the MIMO-OFDM system.Compressed Sensing theory points out that we can use less than the Nyquist rate needed to sampling signal, and the original target signal will be recovered accurately by using reconstruction algorithms if the sampled signal is sparse. The purpose of this paper is to reduce the number of pilots in the process of channel estimation, and saving resources of frequency band without affecting performance of the system. Since most of the MIMO-OFDM channels have sparse or approximately sparse channel parameters, the paper use Compressed Sensing theory instead of the Nyquist theorem to sample channel parameters, and taking Compressed Sensing reconstruction algorithms to recover parameters. In order to insure the performance of the channel estimation and make up for the inadequacy of random pilot, this paper designs a kind of determinate pilot instead of traditional pilot with uniform distribution or random pilot. The simulation results show that, in the algorithm based on Compressed Sensing theory, determinate pilot can keep the same channel estimation performance with random pilot. Pilot with uniform distribution can not get a satisfied estimation result, although it can achieve the best performance in traditional MIMO-OFDM channel estimation algorithms. Finally, the simulation results show that Compressed Sensing theory can dramatically reduce the number of pilots needed to be transferred in the process of MIMO-OFDM channel estimation, meanwhile ensure the estimation performance of high accuracy, to achieve the purpose of the research work in this paper.