Adapting MIMO transmission mode along paths in urban environments

We present algorithms for adapting the MIMO transmission mode to varying channel conditions. Using real paths in Manhattan and Boston, we show that no single fixed transmission mode is superior to others over an entire path. Our two proposed switching algorithms, delayed feedback and probabilistic switching, achieve rates within 3% of the channel capacity. The delayed feedback algorithm requires less switching than the probabilistic switching algorithm. We compare the use of switching with the use of fixed mode transmissions over 50-m long paths. We show that 1) the use of switching rather than the best fixed mode gives at most a 9% gain in average data rate; and 2) the data rate gain relative to using the worst fixed mode can be as high as 90%. A very low rate switching (every 50 m) is thus attainable in theory but it is practical only if the best fixed mode can be accurately predicted.