Task allocation is a typical problem in the area of high performance computing and has been extensively studied in the past. However, existing algorithms cannot be directly used in WSN (wireless sensor network) due to severe energy constraint. A nested optimization technique based on genetic algorithm is proposed for energy-efficient task allocation in multi-hop clusters. The general optimization object can meet application’s real-time requirement while realizing energy efficiency. Optimal solution can be achieved by incorporating GA-based task mapping, GA-based routing, communication scheduling and dynamic voltage scaling (DVS). Performance is evaluated through simulations with randomly generated task graphs and simulation results show better solution in terms of real-time and energy-efficiency compared with random optimization techniques.