The increasing reliance on real-time technologies for building complex systems calls for e.cient dynamic resource management strategies. These strategies aim to achieve high resource utilization and allow graceful degradation in the face of unpredictable workload. In this paper, we present a novel resource management methodology, called Feedback-based Adaptive Resource Management (FARM), which is suited for complex real-time applications that has dynamic workload. We present a case study of one such complex system: Autonomous Hot Spot Convergence System (AHSCS) for sensor web, and apply our resource management solution to AHSCS. The FARM methodology combines the advantages of feedback-controlled scheduling, value-based scheduling, and path-based paradigm to provide a predictable, reliable, and robust services to complex real-time systems, such as AHSCS.