The conventional single-antenna receiver suffers in wireless fading channels from limitations that preclude deployment of envisioned wireless applications. By increasing complexity, improvements are possible using multi-branch receivers. In particular, smart antenna arrays em- ploy maximal-ratio combining (MRC) or statistical beamforming (BF) to exploit diversity and array gain. However, varying azimuth spread creates unfavourable spatial correlation conditions that diminish these gains, while BF and MRC complexity remains constant. On the other hand, adaptive eigen-combining can yield near-optimum performance for more efficient resource us- age. This motivates our study of maximal-ratio eigen-combining (MREC).
We unravel the relationship between MREC, BF, and MRC performance, and evaluate their complexity. Outage and average error probability expressions are derived for MREC assuming perfectly and imperfectly known channel gains. These results are specialized to MRC and BF, as well as to well-accepted pilot-symbol-based channel estimation techniques. In the process, new performance analyses are provided.
Numerical results for typical urban scenarios with variable correlation demonstrate MREC’s advantages. Existing criteria for optimum eigen-mode selection in MREC are reviewed, and a new adaptation approach that accounts for channel condition, algorithm complexity, resource availability, and intended performance level, is proposed and evaluated.
These single- and multi-branch receivers are then evaluated on a field-programmable gate array (FPGA) in terms of symbol-detection performance and resource and power consumption.
MREC flexibility is shown to yield near-optimum performance for half of the hardware and power requirements of MRC, or, equivalently, a doubling of the number of users which can be handled with the same hardware. Smarter antennas, i.e., array receivers aware of the channel- statistics, resource availability, and required performance, can thus be deployed.
Finally, for code-division multiple access (CDMA) systems, we specify an eigen-combining approach. A recently-developed signal despreading method, which eliminates the intended signal, is exploited for interference-plus-noise correlation matrix calculation. After some trans- formations, combining can once again be relegated to a few eigen-modes, for lower complexity and improved performance.