An algorithm for multi-input multi-output (MIMO) adaptive filtering is intr
oduced that distributes the adaptive computation over a set of linearly con
nected computational modules, Each module has an input and an output and tr
ansmits data to and receives data from its nearest neighbor,
A gradient-based algorithm for adapting the parameters in each module to mi
nimize the global mean-squared error is derived using principles of hack pr
opagation, The performance surface is explored to understand the characteri
stics of the adaptive algorithm, The minimum mean-squared error is a many t
o one function of the parameters; therefore, upper bounds on each parameter
are used to prevent excessive parameter drift and insure stability with fi
xed step sizes, Guidelines for choosing the LMS algorithm step sizes and in
itial conditions are developed. Several examples illustrate the performance
of the algorithm.