In this study, the authors propose a cross layer design strategy that consists of a cooperative maximum likelihood detector operating in conjunction with link selection for cooperative multiple-input multiple-output relay networks. The system considered is a two phase relaying model, with amplify-and forward relays. The model considers the effect of relay placement and of large scale shadowing. The authors develop a cooperative maximum likelihood detector for an arbitrary number of relays, and link selection schemes are devised for two scenarios relating to the available knowledge at the destination node, which considers relay combination, with complexity analysis. The authors also propose a cooperative list sphere decoder that processes soft information and implements an iterative detection and decoding scheme. The results of simulations of the systems are presented, with comparisons to system models used in previous literature, and show how the cooperative detector and the proposed link selection schemes affect the bit error rate performance at the destination node, for both the hard decision non-iterative case, and the soft information iterative processing case.