Automotive industry is moving towards a new, highly-scalable and modular embedded architecture developed on component based software and powerful platforms. Moreover increasing complexity will make current approach no longer manageable. Such a system would benefit from the introduction of self-capabilities to enhance performance, safety and life-span. This master thesis presents a self-healing module that enables faults all over the system to be detected, localized and solved using a rule based-engine. Then it is analyzed the limit of a pure centralized approach and a solution is found in the dynamic installation of the self-healing module on different platforms in the network, based on a leader election algorithm.