Discrete Rate Scheduling for Packets With Individual Deadlines in Energy Harvesting Systems
Discrete Rate Scheduling for Packets With Individual Deadlines in Energy Harvesting Systems
复制标题
能量收集系统中具有单独截止日期的数据包的离散速率调度
DOI：
10.1109/jsac.2015.2391491
复制
发表时间：
2015-03-01
影响因子：
16.4
通讯作者：
Shen, Xiaojun
中科院分区：
文献类型：
--
作者：
Shan, Feng;Luo, Junzhou;Shen, Xiaojun
本文提出了一种称为截断的最优速率调度算法，用于能量采集的无线发射机以最小的传输能量传输一组动态到达的分组。与现有的工作不同，我们允许分组具有单独的延迟约束，这是有史以来假设的最通用的模型，但非常需要保证每个应用的服务质量(Qos)。此外，我们将允许的费率限制为一组离散值，这更实用，在许多实际应用中也是需要的。作为第一个成果，我们得到了一个最优的离线算法，它假设速率是连续可调的。然后，我们提出了一个通用的框架，它将任何使用连续速率模型的算法转换为只使用离散速率的算法，同时保持最优性，只要凸速率-幂函数的最优性成立。可能是收获的能量不足以保证所有的包都能在最后期限内完成。如果发生这种情况，以有限的可用能量最大化吞吐量成为要实现的目标。我们的截断算法能够识别这种情况，并在分组共享共同的截止日期的情况下生成保证最大吞吐量的调度。此外，基于最优离线算法，设计了一种高效的在线算法，仿真结果表明该算法能够产生接近最优的结果。
This paper presents an optimal rate scheduling algorithm called Truncation for an energy-harvesting enabled wireless transmitter to transmit a set of dynamically arrived packets with minimum transmission energy. Distinct from existing works, we allow packets to have individual delay constraints, which is the most general model ever assumed but is very much desired to guarantee per-application quality-of-service (QoS). Moreover, we restrict the allowable rates to a set of discrete values, which is more practical and required in many real applications. As the first achievement, we obtain an optimal offline algorithm, which assumes the rate is continuously adjustable. Then, we propose a general framework that transforms any algorithm using the continuous-rate model into an algorithm using only discrete-rates, while preserving the optimality as long as the optimality holds for convex rate-power functions. It is possible that the harvested energy is insufficient to guarantee all packets to meet their deadlines. Should this occur, maximizing throughput with the limited available energy becomes the goal to achieve. Our Truncation algorithm is able to identify this case and produces a schedule that guarantees maximum throughput, if packets share a common deadline. Furthermore, based on the optimal offline algorithms, an efficient online algorithm is designed which has been shown by simulations to produce near optimal results.