We introduce PULSE, a sub-$\mu$s optical circuit switched data centre network architecture controlled by distributed hardware schedulers. PULSE is a flat architecture that uses parallel passive coupler-based broadcast and select networks. We employ a novel transceiver architecture, for dynamic wavelength-timeslot selection, to achieve a reconfiguration time down to O(100 ps), establishing timeslots of O(10 ns). A novel scheduling algorithm that has a clock period of 2.3 ns performs multiple iterations to maximize throughput, wavelength usage and reduce latency, enhancing the overall performance. In order to scale, the single-hop PULSE architecture uses sub-networks that are disjoint by using multiple transceivers for each node in 64 node racks. At the reconfiguration circuit duration (epoch = 120 ns), the scheduling algorithm is shown to achieve up to 93% throughput and 100% wavelength usage of 64 wavelengths, incurring an average latency that ranges from 0.7–1.2 $\mu$s with best-case 0.4 $\mu$s median and 5 $\mu$s tail latency, limited by the timeslot (20 ns) and epoch size (120 ns). We show how the 4096-node PULSE architecture allows up to 260 k optical channels to be re-used across sub-networks achieving a capacity of 25.6 Pbps with an energy consumption of 82 pJ/bit when using coherent receiver.

中文翻译：

---

PULSE：以纳秒时间尺度运行的光电路交换数据中心架构

我们介绍 PULSE，一个子$\mu$由分布式硬件调度器控制的光电路交换数据中心网络架构。PULSE 是一种扁平架构，它使用基于并行无源耦合器的广播和选择网络。我们采用了一种新颖的收发器架构，用于动态波长时隙选择，以实现低至 O(100 ps) 的重新配置时间，建立 O(10 ns) 的时隙。一种时钟周期为 2.3 ns 的新型调度算法执行多次迭代，以最大限度地提高吞吐量、波长使用率并减少延迟，从而提高整体性能。为了扩展，单跳 PULSE 架构使用不相交的子网络，通过为 64 个节点机架中的每个节点使用多个收发器。在重配置电路持续时间（时代 = 120 ns)，调度算法显示可实现高达 93% 的吞吐量和 64 个波长的 100% 波长使用率，导致平均延迟范围为 0.7-1.2 $\mu$s 与最佳情况 0.4 $\mu$s 中位数和 5 $\mu$s 尾延迟，受时隙 (20 ns) 和纪元大小 (120 ns) 限制。我们展示了 4096 节点 PULSE 架构如何允许在使用相干接收器时跨子网重复使用多达 260 k 个光信道，实现 25.6 Pbps 的容量和 82 pJ/bit 的能耗。