The use of arbitrary sequences for real-time wireless channel evaluation

The need for accurate channel measurements for broadband communication systems is well known. The problem of system identification constitutes the basis for any wireless channel measurement and estimation. The classical approach to system identification employs a pseudo-random (PR) sequence as a test signal. The system response to this PR sequence is then correlated with a replica of the test signal in order to obtain an estimate of the system impulse response. The fundamental assumption upon which this technique is based is that the PR test signal has an ideally impulsive autocorrelation function (ACF). In practice, the majority of sequences of interest do not possess this idealised ACF. The consequences of employing test signals with nonimpulsive ACF results in the introduction of estimation errors. The problem is further accentuated when the classical approach to system identification is applied to channel estimation. In this paper, the problem of system identification is analysed for the case when the ACF of the test signal is non-ideal. It is shown that the system impulse response estimation error can be reduced by using a compensation filter whose response has been related to the imperfection in the test signal ACF. The applicability of this approach to broadband channel estimation is then demonstrated.