FBMC'de İç-girişim ve Öncül Tabanlı Kanal Kestirimi

Abstract

Multi-carrier communication systems perform well in the systems they are used. Therefore, it is considered to be used in the next generation communication systems. Among multi-carrier systems, OFDM is the most preferred method. The most important reasons for common use of OFDM are low complexity and high spectral efficiency. On the other hand, some features such as high power side-lobs and high PAPR causes to search for an alternative method to OFDM for 5G.

A candidate method for 5G is OFDM/OQAM, which is an FBMC member. Although OFDM/OQAM provides better performance than OFDM in many respects, because of intrinsic-interference seen in the receiver due to the system structure, conventional channel estimation methods can not be applied directly. For this reason, new channel estimation methods have been developed for OFDM/OQAM.

The main purpose of this thesis is to examine the general FBMC structure and particularly the studies related to the OFDM/OQAM in the literature, then to understand the intrinsic-interference which is the specific feature of OFDM/OQAM, and finally the most popular preamble based channel estimation methods for OFDM/OQAM and their comparison.

For this purpose, firstly the FBMC concept and then the basic OFDM/OQAM structure are described. The theory of Nyquist channel, which enables communication without ISI in the system, is explained and then the PHYDYAS function, which is used to obtain Nyquist channel in OFDM/OQAM, is investigated. The OFDM/OQAM signal and intrinsic-interference in the receiver is shown mathematically. After that, instead of getting rid of intrinsic interference, utilization of it and its contribution to channel estimation and information capacity of the channel are investigated. It was shown that a capacity-based gain of over 2.5 dB can be achieved via utilization of intrinsic-interference.

Finally, OFDM/OQAM channel estimation methods POP, ICM, IAM-1, IAM-R, IAM-I, IAM-C and E-IAM-C are examined. Their BER performances in the presence of AWGN, only on Rayleigh and both Rayleigh and Doppler channel are shown. The influence of using high-order constellation diagrams on methods is examined. The methods are also compared in terms of their PAPR performances.

As a result of comparisons, it was found that POP is a suitable method for high-order constellation diagrams, even though it shows the worst BER performance for low-order constellation diagrams. Although E-IAM-C provides the lowest bit error probability against high noise power, it shows poor performance in terms of PAPR. IAM-1 is the best method in terms of PAPR and the worst method in terms of BER, the best method provides PAPR-BER balance was determined as IAM-I. In addition, an ICM preamble for the PHYDYAS function was designed in this thesis, and the simulations shows that the method has balanced PAPR and BER performance and given unsatisfactory results for high-order constellation diagrams .