Abstract
An adaptive channel shortening equalizer design for multiple input multiple output-orthogonal frequency division multiplexing (MIMO-OFDM) radio receivers is considered in this presentation. The proposed receiver has desirable features for cognitive and software defined radio implementations. It consists of two sections: MIMO decision feedback equalizer (MIMO-DFE) and adaptive multiple Viterbi detection. In MIMO-DFE section, a complete modified Gram-Schmidt orthogonalization of multichannel input data is accomplished using sequential processing multichannel Givens lattice stages, so that a Vertical Bell Laboratories Layered Space Time (V-BLAST) type MIMO-DFE is realized at the front-end section of the channel shortening equalizer. Matrix operations, a major bottleneck for receiver operations, are accordingly avoided, and only scalar operations are used. A highly modular and regular radio receiver architecture that has a suitable structure for digital signal processing (DSP) chip and field programable gate array (FPGA) implementations, which are important for software defined radio realizations, is achieved. The MIMO-DFE section of the proposed receiver can also be reconfigured for spectrum sensing and positioning functions, which are important tasks for cognitive radio applications. In connection with adaptive multiple Viterbi detection section, a systolic array implementation for each channel is performed so that a receiver architecture with high computational concurrency is attained. The total computational complexity is given in terms of equalizer and desired response filter lengths, alphabet size, and number of antennas. The performance of the proposed receiver is presented for two-channel case by means of mean squared error (MSE) and probability of error evaluations, which are conducted for time-invariant and time-variant channel conditions, orthogonal and nonorthogonal transmissions, and two different modulation schemes.
Highlights
The fundamental problem in the design of future wireless communication systems is to reliably and efficiently transmit and receive information signals over imperfect channels using substantially high data rates
One successful approach adopted in several wireless standards such as digital audio broadcasting (DAB), digital video broadcasting (DVB-T), local area networking (LAN), and metropolitan area networking (MAN) is orthogonal frequency division multiplexing (OFDM) in which the entire bandwidth is divided into several narrow subbands so that the frequency response over each individual subband is relatively flat, and each subband channel occupies only a small fraction of the original bandwidth
We focus on the reception mode of operation of cognitive multiple input multiple output (MIMO)-OFDM radios and propose a new minimum mean squared error (MSE) channel shortening equalizer design, which consists of adaptive fron-tend MIMO-DFE and multiple Viterbi detection sections
Summary
The performance achieved through MIMO technology entails a considerable increase in signal processing complexity in receiver, and there exists a major challenge in designing low-complexity receivers for multichannel wireless systems. 2.1 V-BLAST type MIMO-DFE We would like to use a V-BLAST type design approach for the front-end filter of the proposed equalizer, and we require the design of a single, multichannel, and compact equalizer structure, so that two separate equalizers and direct evaluations as in (16) are avoided, and the same filter can be reconfigured as spectral analysis or positioning filter These objectives can be accomplished by considering the equivalence of V-BLAST and modified Gram-Schmidt orthogonalization operations, and by completely orthogonalizing the two-channel input data of DFE using SPMGLSs, which provide scalar only operations, good numerical properties as well as modularity, regularity, order recursiveness, and reconfigurability to the solution of equalization problem under consideration. Similar to two-channel and three-channel cases, we solve the optimization problem in (43) using the new data vector in Equation (46), in which case θ (n) and θ (n) are 4θ × 4θ lower triangular transformation, and 4 × 4θ joint process estimation error coefficient matrices at the time instant n, respectively. These matrices are computed stage-by-stage by the use of 4 × 4 lower triangular transformation matrices, Lθ (n), and 4×4 joint process estimation error coefficient matrices, θ (n), at time instant n, respectively
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
