Optimal training sequences for TDMA systems

This paper deals with the problem of finding optimal training sequences for adaptive equalizers in TDMA systems. Such sequences give the equalizer a good initial value, are required for symbol-timing recovery and can provide a small but robust amount of information by utilizing a small set of codes with strong discrimination power. While every single task has been dealt with in literature, the combination of all three of them is a new problem that requires smart search strategies to explore a huge space of possibilities. On the example of a local wireless loop design, it is shown how the search problem can be reduced.