In this paper, we investigate the problem of distributed computing in tactical wireless sensor networks (WSNs). The constituent nodes of such WSNs are individually resource poor, but are collectively capable of carrying out significant calculations, if appropriately orchestrated. We propose a general distributed wireless ad hoc grids (DWAG) paradigm, exploiting concepts from the field of computational grids and modifying them in order to respect the constraints inherent in ad hoc sensor networking. In particular, we use our proposed paradigm to implement the CCA-MAP localisation algorithm on a real WSN test-bed. We investigate the impact of radio interference on the performance of the system, and propose a technique to ameliorate its effects. Our experimental results, conducted using different network topologies, demonstrate that it is possible to achieve significant performance improvements in terms of job latency by considering both the computing load and the network constraints.