Abstract:
Wide signal bandwidths typically require receivers performing signal detection to collect very
large quantities of data. However, in applications with limited size, weight and power (SWAP)
requirements, reducing the amount of data becomes important for proper operation. Most
existing sample reduction approaches rely on reconstruction algorithms to compensate for the
missing data, but these are often computationally complex. Therefore, in this work sample
reduction without reconstruction is considered.
This work proposes an approach to discarding samples prior to detection using difference sets
(DSs) and almost difference sets (ADSs) – exploiting their sidelobe and cyclic properties – to
minimise the negative impact on detection performance. Included are mathematical analyses,
simulations, and experiments with practical data evaluating the effects of this technique on
the detection performance. This work demonstrates that while the lack of a reconstruction
algorithm does introduce interference, this is reduced when using DSs and ADSs compared
to when samples are discarded at random, and the use of these sets allows predictions about
performance to be made beforehand using only the set parameters. Additionally, the proposed
technique performs much faster than detection with reconstruction, while having a reasonable
decrease in detection performance.
