Abstract: For solving the inconsideration of power issues in current application deployment decision-making system, a more thoughtful system model is proposed by adding the power factor into the model. What’s more, a parameter-adjustable windowed mapping algorithm is designed to deal with the problem when the number of processors that composes the network in SDR platform is relatively high. The simulation reveals that by applying the proposed model, instead of current system model, total energy can be saved by nearly five percentages and with the proposed algorithm, the performances and complexity of algorithm can be balanced so that user-tolerable performances can be achieved and executing time can be saved by tuning the parameter.

Key words: Software Defined Radio（SDR）, computation resource management, application deployment decision-making, pipelined scheduling, back-track algorithm

摘要： 针对当前软件无线电（Software Defined Radio，SDR）应用部署决策系统的应用模型在功率消耗方面考虑的不足，以及现有应用部署算法在处理器数目较大时复杂度高的问题，提出了更为完善的系统模型，并在该模型的基础上提出参数可调整的加窗算法。仿真结果表明，同现有系统模型相比，提出的模型能够保证服务质量，可将系统功耗降低约5%，同时提出的算法兼顾复杂度和性能，可通过调整参数，适当降低复杂度并获得用户可容忍的性能。

关键词: 软件无线电, 计算资源管理, 应用部署决策, 流水线调度, 回溯法