As the Linux kernel continues to gain features which permit it to operate with lower scheduling latencies, it is becoming increasingly feasible to utilise systems running stock kernels for high speed data acquisition in applications such as radar. This makes such applications much more accessible to both developers and end users. However, having low scheduling latency is not enough to ensure smooth operation of the system. The acquisition software itself must also be carefully designed to ensure it can keep up with the data as it arrives and distribute it to analysis processes.

After briefly describing the needs of a radar system, this talk will outline the various programming approaches and techniques utilised in the implementation of a radar data acquisition system and explore the reasons behind the choices made. The presentation will be aimed at developers who have not necessarily worked with data acquisition before.