Multichannel Particle Filters for Tracking A Frequency Hopped Signal

We consider the problem of tracking a frequency-hopped signal without knowledge of its hopping pattern. The problem is of interest in military communications, where, in addition to frequency, hop timing may also be randomly shifted to guard against unauthorized reception and jamming. We have recently proposed (Sidiropoulos et al., 2006) a conceptually simple stochastic state-space model that captures the randomness in carrier frequency and hop timing, and is well- suited for application of particle filtering tools. In this contribution, we generalize the approach in (Sidiropoulos et al., 2006) to the multichannel case. In particular, we show that the importance function that minimizes the variance of particle weights can again be computed in closed form. The resulting multichannel particle filters are assessed and compared to their single-channel counterparts in pertinent simulation experiments.