Pilot Assisted and Semiblind Channel Estimation for Interleaved Frequency Division Multiple Access and the Generalization for Block-Interleaved Frequency Division Multiple Access

For the uplink of future mobile radio systems, Interleaved Frequency Division Multiple Access (IFDMA) is a promising multiple access scheme as, in comparison to other multiple access schemes, IFDMA offers several advantageous properties that are of special interest for uplink transmission. For example, due to the precoding of the data symbols and the allocation of distributed subcarriers to a user under consideration, the IFDMA transmission is most reliable as it provides high frequency diversity. Further, the IFDMA transmit signal exhibits a low Peak-to-Average Power Ratio (PAPR) which enables the utilization of cost-efficient amplifiers in the mobile terminals. A generalization of the IFDMA scheme is the Block-Interleaved Frequency Division Multiple Access (B-IFDMA) scheme, where the precoded data symbols are transmitted on distributed blocks of neighboring subcarriers. The B-IFDMA scheme maintains the advantages of IFDMA and, at the same time, exhibits less sensitivity against carrier frequency offsets than IFDMA. One major demand in order to make use of the reliable data transmission provided by IFDMA and B-IFDMA is the estimation of the mobile radio channel with high accuracy and low pilot symbol overhead. Channel estimation for the uplink transmission with IFDMA and B-IFDMA entails new challenges as precoded data symbols are transmitted on distinct subcarriers that are distributed over the whole available bandwidth. This project introduces new channel estimation algorithms for IFDMA as well as B-IFDMA.

First, the channel is estimated for IFDMA with the help of pilot symbols that are inserted into the data stream and evaluated at the receiver. Two different pilot insertion methods and corresponding estimation algorithms are introduced under consideration of maintaining the low PAPR of the IFDMA transmit signal and providing reliable estimation performance while consuming a low pilot symbol overhead. It is revealed that pilot assisted channel estimation for IFDMA exhibits good estimation performance but suffers from a high pilot symbol overhead. The high pilot symbol overhead is due to the transmission on distributed subcarriers in frequency domain which prevents the application of interpolation filtering in most cases.

In order to avoid the high pilot symbol overhead for IFDMA channel estimation, in this project, also semiblind channel estimation is investigated for IFDMA. The semiblind channel estimation exploits the information about the channel that is inherent to the received data signal. Thus, the application of semiblind channel estimation allows to reduce the pilot symbol overhead compared to pilot assisted channel estimation. For the semiblind estimation of frequency domain channel variations, two algorithms are introduced that take advantage of the cyclostationarity of the IFDMA signal. In order to additionally estimate the channel variations in time domain, these two algorithms are combined with a so-called decision directed estimation approach which utilizes the received data symbols as virtual pilot symbols. The analysis of the proposed semiblind channel estimation algorithms confirms their benefit in terms of pilot symbol overhead reduction compared to the introduced pilot assisted channel estimation approaches.

Furthermore, it is shown that the estimation performance of semiblind channel estimation is comparable to the estimation performance of pilot assisted channel estimation in case of a low noise power.

Finally, channel estimation for B-IFDMA is addressed. As B-IFDMA represents a generalization of IFDMA, the pilot assisted and semiblind channel estimation algorithms derived for IFDMA are investigated with respect to their applicability to B-IFDMA.

It turns out that, due to the transmission on distributed blocks of subcarriers, B-IFDMA opens up the possibility of interpolation in frequency domain and, thus, supports the derivation of new pilot insertion methods and estimation algorithms compared to IFDMA. Further, it is revealed that semiblind channel estimation is also applicable to B-IFDMA and entails a beneficial reduction of the pilot symbol overhead.