Sparse time-variant impulse response estimation for vehicular channels using the c-LASSO

Vehicle-to-Vehicle channels pose large modeling challenges due to their time-varying nature and their large Doppler shifts. While new descriptions for the channel are being researched, many standardization bodies and hardware manufacturers still rely on the concept of a tapped delay line as channel description. In this paper, we rephrase the problem of estimating a tapped delay line with given number of taps as sparse array processing problem. We formulate the problem via the complex LASSO algorithm, and modify the cost function to only consider frequency regions of interest. We thus end up at a description that estimates a tapped delay line impulse response from channel sounder data in such a way that a given subspace of frequencies is optimally fitted. Our results show that this combination of subspace projection and LASSO optimization allow to estimate the desired channel of interest with high precision and provides large gains over conventional delay line estimation methods.