Prediction and Modeling for the Time-Evolving Ultra-Wideband Channel

We conduct a feasibility study of ultra-wideband (UWB) channel prediction to answer the following two questions: Is the UWB channel predictable? Is UWB channel prediction useful? We setup the problem in the following way: A receiver travels along a linear trajectory at a constant velocity. The transmitter and environment are stationary. Using past channel measurements, the receiver predicts future measurements of the channel, assuming its direction of movement and velocity remain constant. Our approach is to decompose the time evolution of the channel, which is jointly correlated in time and delay, in terms of the time evolution of individual paths, which are independent across delay. A measurement campaign was conducted in the Berkeley Wireless Research Center, where measurements were taken with line-of-sight (LOS) and non-line-of-sight (NLOS) conditions. We develop a channel prediction algorithm, and evaluate results in terms of the matched filter output energy (MFOE). Iterating through the six strongest paths, our prediction algorithm achieves more than 70% (40%) of the possible MFOE over a prediction distance of 34 cm for the LOS (NLOS) conditions. These results are good since the coherence distance, being the distance for which the channel is approximately constant, is less than 1 cm.