In this paper, a DSP-based solution has been proposed from control-theoretic paradigm to control the congestion of a multi-hop ad hoc wireless LAN with bandwith-on-demand access. The design of the proposed system has been derived from the reference and error controller model, which essentially controls the queue length against a reference queue length based on the congestion that the system has to combat against. This is done by regulating the desired input rate and required output rate such that the system is asymptotically stable in terms of all possible congestion scenarios. The Smith predictor has been used in the closed-loop error controller to compensate for delays that could cause instability. Underpinning the control paradigm, a filter based solution is proposed with an aim to ensure full-link utilization and to achieve maximum rate recovery as soon as the congestion has been cleared under system stability. Simulations illustrate the system performance of the filter-based controller in a simple WLAN model. Finally, a brief comparitive study has been made with some of the recently proposed models that perform similar control tasks.