Past Research: Non-coherent Communication using Roots of Polynomials
Most wireless communication links are coherent, meaning the receiver (and often the transmitter, too) use knowledge of the channel to enable more reliable communication. But acquiring knowledge about the channel is not free: it requires overhead signaling (e.g., pilot signals), which reduces the throughput and wastes energy. Nevertheless, such overhead is generally tolerated, as it is sparse relative to the data and improves the overall system performance.
For short-packet transmission, however, reference overhead can comprise a non-negligible fraction of the total packet and even dominate the information exchange. In this regime, an attractive alternative is non-coherent communication, where the received signal is processed without knowledge of the instantaneous channel state information (no pilots are needed).
This research studied a novel non-coherent technique called binary modulation on conjugate-reciprocal zeros (BMOCZ). The principle of BMOCZ is quite unique: transmit the coefficients of a polynomial whose zeros (i.e., roots) encode the bits. With this approach, convolution corresponds to polynomial multiplication, which preserves the transmitted data zeros, regardless of the channel realization!
Together with my coauthors, I proposed signal processing techniques for BMOCZ to yield robust performance under hardware impairments at the physical layer (e.g., time and frequency offsets). I also analyzed BMOCZ in more conventional communication scenarios, such as uplink OFDMA, where I exploited its autocorrelation properties to design a pilot-free, fixed-PAPR, OFDM-BMOCZ framework.
The bulk of this work is presented in my paper available here; I also have a related GitHub repo here.
