Rapidly varying sparse channel tracking for OFDM systems

Abstract

Over the 30 or 40 years wireless communication has got too much attention in the communication field mainly because of the mobility requirements. Therefore ongoing researches focused to improve performance of the wireless communication systems which consist of receiver and transmitter part. Channel impairments are the main detrimental effects for communication systems since transmitted signal passed through the channel. Receiver is tried to detect message signal from received corrupted transmitted signal. Also, additive channel noise and communication system noise decrease the performance of detection the message signal. Wireless communication channel causes the interference between symbols and channels because of the fading. Thus, wireless communication systems are designed that capable to mitigate channel effects. Based on practical observations some of the researchers are modeling the channel distribution for considered environments. Some of the observation based channel models in literature are Nakagami, urban and rural channel models. Also, there is other different methods to model the channel. Channel equalization is important for reliable communication between transmitter to receiver. Therefore many researchers are aimed to estimate the channel. These studies basically grouped as parametric and blind channel estimation methods. Parametric means that channel properties are parameterized depend on the specified features of the communication environment. There is many approaches to model the channel in the literature. On the other hand, blind channel estimation methods are not use the assumptions therefore its performance is less than parametric methods. Blind methods are more convenient for situations where there is no channel information. One of the main restrictions of the communication systems is bandwidth. Due to available bandwidth is limited, researchers are led to more spectral efficient modulations. Orthogonal Frequency Division Multiplexing (OFDM) technique is widely accepted by many communication standards because of the spectral efficiency. OFDM divides available bandwidth to orthogonal subchannel and transmits high data rate message signal via parallel slow data rate subcarrier. When the channel is slow fading, subchannel of the OFDM system will be flat. Therefore the channel estimation and equalization will be handled more easily. However, when the channel is fast varying the orthogonality between sub channels will be violated. Thus, inter carrier interference will occur, one OFDM symbol will face with many channel coefficients which changing over time depend on velocity. There is many studies and approaches to increase performance of the OFDM system while the channel is fast varying. It is important to use resources of the communication system effectively. So, recent studies are trying to exploit sparsity information which enables better performance even insufficient observation. Recent compressed sensing studies showed that sparse signals can be recovered even there is not enough observation. Even these algorithms are rather complex, some of them can be applied to the channel estimation problems. In literature it is accepted that the wireless communication channel has sparse nature. It is generally assumed that the channel length is limited and only strong paths will have effect on signal. Sparse channel means that there is only few multi path over the channel length. In this thesis we will focus on parametric and data aided channel estimation problem for OFDM systems. Our subject mainly based on Autoregressive (AR) model which use velocity of the mobile receiver and sparsity assumption. Under these assumptions, SAGE MAP based channel estimation method in this thesis. Each section will be explained as follows. Introduction section will present general view to wireless communication system. The studies examined which focus on channel estimation. Proposed channel models and channel estimation algorithms in the literature is explained. Literature review about CS approach for channel estimation is summarized. Also sparse and fast varying channel estimation studies are explained briefly. And the main objective and the aimed contributions of the thesis are explained. Section of wireless communication channels gives descriptions of the channels and summarize channel modeling parameters. Also generally accepted channel models are explained. Third section explains the OFDM system and its advantages, disadvantages. OFDM block scheme is presented and process flow is explained. The OFDM channel estimation section outlines fundamental methods to estimate channel for OFDM signals. Also simulation results included to analyze and compare these methods with each other. the channel estimation methods explained in this section do not consider the channel sparsity. Therefore their performances can be improved with CS algorithms. Recent years it is proved that many real world channels are naturally sparse. Sparsity knowledge which is a priori assumption about channel, can improve the estimation performance. Therefore fourth section outlines definition of sparse signal, sparse channel, compressed sensing techniques and sparse channel estimation. Moreover, simulation results are obtained to compare proposed method pros and cons. Fifth section clarify the sparse fast varying channel which time variation of the channel is modeled with AR process. Then proposed problem definition is presented and explained with details. Derivation of the problem is explained and at the end we concluded with an algorithm. Simulation of the algorithm is presented and compared with other case studies and analyzed. In conclusion, it is explained that how proposed algorithm can be extended and how its weaknesses can be eliminated. It is showed that proposed method is more flexible to track channel sparsity. Therefore it can be useful for future communication standards which requires to acceptable communication performance under fast varying environment.