Pilot Signal Design For Estimation Of Sparse Channels With Application To Cooperative Systems

Abstract

In recent years, there has been a resurgence in the interest of using HF radio band (3-30 MHz) for military, government, and emergency applications. In order to allow for more reliable mobile HF radio networks, the use of distributed cooperative communication schemes is appealing. In fact, the large HF wavelength prevents one from obtaining diversity from antennae located on a single portable radio. Hence cooperative schemes can potentially bring the benefits of MIMO systems to portable HF radios, allowing for diversity gains as well as potential reductions in transmit power. Our work is focused on the receiver design for cooperative HF radios. The motivation for our work is as follows. Using distributed cooperative communication schemes in OTH-HF propagation environments introduces several additional complications in comparison with traditional MIMO systems. Different from a traditional MIMO system, there is additional overhead required to organize and synchronize the cooperative radios' transmissions. Our work focuses on the distributed cooperative communication scheme introduced in [82] which reduces the overhead by introducing randomization into the coding scheme, but does not solve the synchronization problem. In terms of efficiency, it is best to allow cooperative nodes to transmit concurrently, using space-time coding techniques, as done in the randomized protocol proposed in [82]. But several authors argue against such solutions, given the challenge of synchronizing the different cooperative radios participating in the transmission. Our work investigates the effects of imperfect synchronization among cooperative nodes, resulting in dispersive effects that can be captured by an equivalent MIMO channel. The key contribution is in proposing the use of compressed sensing (or sparse signal recovery) techniques to deal with these channels at the receiver. We first consider designs that are sufficiently informative for MIMO systems that can be appropriately modeled as sparse. After introducing noise and modeling error, we examine the performance of an ideal sparse channel estimation method, leading to a metric we call localized coherence, and modified training designs. These results are then applied to the asynchronous distributed cooperative communication scheme.