PULS: Processor-Supported Ultra-Low Latency Scheduling
PULS: Processor-Supported Ultra-Low Latency Scheduling
复制标题
DOI：
10.1145/3209582.3209609
复制
发表时间：
2018-06
期刊：
影响因子：
--
通讯作者：
Simon Yau;Ping-Chun Hsieh;Rajarshi Bhattacharyya;K. Bhargav;S. Shakkottai;I.-Hong Hou;P. Kumar
中科院分区：
文献类型：
--
作者：
Simon Yau;Ping-Chun Hsieh;Rajarshi Bhattacharyya;K. Bhargav;S. Shakkottai;I.-Hong Hou;P. Kumar
下一代无线网络支持的越来越多的应用程序要求数据包在特定期限之前到达，以使系统具有所需的性能。虽然已经提出了许多时间敏感的调度协议，但很少经过实验评估来建立实际的性能。此外，其中一些协议涉及需要在每个数据包的基础上执行的高复杂性算法。这些协议的实验评估需要一个灵活的平台，能够轻松地实现和试验这些协议。我们推出了 PULS，这是一种处理器支持的超低延迟调度实现，用于测试具有超低延迟要求的下行链路调度协议。基于我们的解耦架构，延迟敏感调度协议的可编程性在主机上完成，低延迟机制部署在硬件上。这使得软件调度策略灵活，硬件功能复用性高，同时满足MAC的时序要求。我们在该平台上进行了广泛的测试，以验证每个数据包调度所经历的延迟，并提供结果表明数据包可以在 PULS 中调度并在 1 毫秒内传输。使用 PULS，我们实施了四种不同的调度策略，并提供了各种流量负载和实时要求下的详细性能比较。我们表明，在某些场景下，最优策略可以使有期限的数据包的丢失率保持在 1% 以下，而其他协议的丢失率高达 65%。
An increasing number of applications that will be supported by next generation wireless networks require packets to arrive before a certain deadline for the system to have the desired performance. While many time-sensitive scheduling protocols have been proposed, few have been experimentally evaluated to establish realistic performance. Furthermore, some of these protocols involve high complexity algorithms that need to be performed on a per-packet basis. Experimental evaluation of these protocols requires a flexible platform that is readily capable of implementing and experimenting with these protocols. We present PULS, a processor-supported ultra low latency scheduling implementation for testing of downlink scheduling protocols with ultra-low latency requirements. Based on our decoupling architecture, programmability of delay sensitive scheduling protocols is done on a host machine, with low latency mechanisms being deployed on hardware. This enables flexible scheduling policies on software and high hardware function re-usability, while meeting the timing requirements of a MAC. We performed extensive tests on the platform to verify the latencies experienced for per packet scheduling, and present results that show packets can be scheduled and transmitted under 1 ms in PULS. Using PULS, we implemented four different scheduling policies and provide detailed performance comparisons under various traffic loads and real-time requirements. We show that in certain scenarios, the optimal policy can maintain a loss ratio of less than 1% for packets with deadlines, while other protocols experience loss ratios of up to 65%.