Squeezing the spectral lemon: Advances in signal processing and coding to improve spectral efficiency

Over the last few years, a research program was undertaken in collaboration with InterDigital Canada under the theme “Advanced Broadband Transceivers.” The aim was the development of novel signal processing techniques, exploiting the latest in communication theory and having fundamental advantages, while keeping a focus throughout on associated implementation issues, such as algorithmic complexity, power consumption, etc. This research approach, bridging theory and implementation under a pragmatic mindset, yielded rather promising results. For example, it is known that MIMO techniques are the key to augmenting effective link throughput without bandwidth expansion. However, it is still problematic in practice to incorporate multiple antennas on handsets because of cost/power/size constraints. As a solution, virtual MIMO techniques based on sphere decoding were developed which can effectively at the receiver separate more co-channel signals than there are receive antennas. Also, new powerful quasi-cyclic LDPC codes were devised which allow encoding in linear complexity. Furthermore, new joint decoding techniques were developed as well as efficient parallel hardware implementations. Together, these techniques are capable of aggregate throughputs above 10 Gbps on an FPGA. Multi-rate codec architectures were also developed and then applied in ARQ (automated repeat request) schemes, as well as relaying / network coding scenarios. We will look at how these techniques can improve spectral efficiency in current wireless systems. Approaches for further gains will also be discussed, including distributed arrays, cognitive radio, cognitive networks, interference alignment, and network coding.