A large ad hoc network (e.g., in ground disaster area) requires effective fault management and self-healing mechanism to ensure effective information exchange. To achieve this goal, we study a Quality of Service (QoS) - based network fault management and self-healing protocol for network fault detection and self-healing for ad hoc networks with low false detection and network management overhead. The proposed software module addresses several critical limitations of current works for fault management that can be used for detecting and restoring soft failures in the heterogeneous and large-scale networks. The QoS parameters are defined to quantify the analysis of soft failures, and then the self-healing problem in multiple ad hoc domains is formulated as an optimization problem and mathematical models are developed to formulate the problem. Based on the analysis, heuristic self-healing algorithms are proposed for fault restoration across multiple domains to achieve efficiency and scalability in addition to server survivability. The self-healing algorithms enable dynamic domain reconfiguration in response to soft-failure in an effective way.