Priority First Spherical Lattice Space-Time Decoder with Boundary Control

Stack-based sequential priority first decoder for spherical LAST codes with boundary control

A stack-based sequential priority first decoder that returns Maximum-Likelihood solutions to spherical LAST coded MIMO system-type problems, i.e., a lattice decoder with spherical boundary control. In such problems, the depth of the search tree is known but the number of children per node is not fixed. The number of children and their associated symbol decisions are governed by the structure of the spherical LAST code. LAST_dec enforces boundary control by decomposing the code generator matrix in a manner that is like a dual to the standard sequential decoding procedure itself.

Note that inputs to LAST_dec must be real; in practice, this implementation of the sequential decoding algorithm is near-ML because it operates with finite memory. If that memory is exceeded, nodes are dropped from the stack and the number of such dropped nodes is returned.

In addition, this implementation allows specification of the size of the finite memory block (in terms of number of nodes) allocated for its execution. Generally we find that restricting the sequential decoder to finite memory is not a major consideration, as very near-ML performance can be achieved with relatively low allocations.

Please see the function pre-amble (type `help LAST_dec` at the Matlab prompt) for more details.