AG Kommunikationstheorie


Thema:

Graph-based iterative joint detection and channel estimation in OFDM systems

Abstract:

Driven by the demand of multimedia services, modern wireless communication systems require high-bit-rate transmission. One way to achieve higher data rate is to reduce the symbol duration. While the symbol duration is shortened, the multi-path channel will cause severe inter-symbol interference (ISI). To minimize ISI, multicarrier transmission is often used. In multi-carrier transmission, the high-bit-rate data stream is divided into several parallel lower bit-rate streams, and modulated to separate carriers, often called subcarriers. Orthogonal frequency-division multiplexing (OFDM) is a spectrally efficient version of multicarrier modulation, where the subcarriers are selected such that they are all orthogonal to each other. Additionally, a cyclic prefix is employed to completely avoid ISI.

Accurate channel estimation can be used in OFDM systems to improve the performance by allowing for coherent demodulation. Also the knowledge of the channel state information can help the transmitter to determine the optimal resource allocation strategy. OFDM convert the frequency-selective fading channel to a series of frequency flat fading subchannels. Thus, channel estimation can be done in frequency domain, where the channel coefficients of each subchannel are estimated individually. Traditionally, the power or amount of training symbols are increased to improve the channel estimation quality. However, doing so will degrade either the power efficiency or the bandwidth efficiency of the system. In comparison, semiblind channel estimation (SBCE) tries to extract the channel state information carried by all observations, and is able to achieve very low mean square error (MSE) with a small number of training symbols. Furthermore, instead of using hard decision of channel coefficients and data symbols, the soft information of data symbols and the uncertainty of channel estimates should be utilized in the iterative process.

Factor graph is a bipartite graph visualizing the factorization of certain global functions subject to minimization or maximization. It is often helpful in design of low-complexity iterative processing algorithms. By treating channel coefficients as variables as well as data symbols, data detection and channel estimation can be performed over a general factor graph. Differed from the conventional schemes, in which data detection and channel estimation are performed in separate manner, this algorithm does everything in one stage.



zurück zur Terminübersicht