Abstract:Due to limited energy supply of Internet of Things (Zhao et al. 2018) (IoT) devices, asynchronous duty cycle radio management is widely adopted to save energy. Since the sleep schedules of nodes are unsynchronized, a sender has to repeatedly send frames to coordinate with its receiver or keep sleeping until the receiver's wake-up time will come according to receiver's sleep-wake schedule. In such contexts, opportunistic forwarding, which takes the earliest forwarding opportunity instead of a deterministic forwarder, shows great advantage in utilizing channel resource for duty-cycled IoT networks. The multiple forwarding choices with temporal and spatial diversity increase the chance of collision tolerance in opportunistic forwarding, potentially enhancing the overall performance of duty-cycled multi-hop networks. However, since the current channel contention mechanisms mainly focus on collision avoidance, it is too conservative to exploit concurrency. To address this problem, in this article, we propose COF to fully exploit the potential Concurrency for Opportunistic Forwarding in duty-cycled IoT networks. COF achieves concurrent transmission by: (i) measuring conditional link quality under the interference of on-going transmissions, and then (ii) further modeling the benefit of potential concurrency opportunities. According to the expected benefit of concurrency, COF decides whether or not to transmit in concurrent way. COF also adopts concurrency flag and signal features to avoid data collision caused by disordered concurrent transmissions and enhance the accuracy of conditional link quality estimation. COF can be easily integrated into the conventional unsynchronized and duty-cycled protocols. We have implemented COF and evaluated its performance on a 40-node testbed. The results show that COF can effectively exploit potential concurrency in opportunistic forwarding and COF outperforms the state-of-art protocols under diverse traffic load and network density.

Django: Bilateral coflow scheduling with predictive concurrent connections

Exploiting Concurrency for Opportunistic Forwarding in Duty-Cycled IoT Networks

US-Byte: an Efficient Communication Framework for Scheduling Unequal-Sized Tensor Blocks in Distributed Deep Learning

Distributed Bottleneck-Aware Coflow Scheduling in Data Centers

Joint Online Coflow Routing and Scheduling in Data Center Networks

Cross-Layer Self-Similar Coflow Scheduling for Machine Learning Clusters.

Coflow Scheduling in Data Centers: Routing and Bandwidth Allocation

Efficient Scheduling for Multi-Stage Coflows

A Deadline-Aware Coflow Scheduling Approach for Big Data Applications.

Coflow Scheduling in Input-Queued Switches: Optimal Delay Scaling and Algorithms

DCoflow: Deadline-Aware Scheduling Algorithm for Coflows in Datacenter Networks

Efficient Scheduling of Weighted Coflows in Data Centers

Endpoint-Flexible Coflow Scheduling Across Geo-Distributed Datacenters

Coflow Scheduling in the Multi-Resource Environment

Semi-Distributed Coflow Scheduling in Datacenters

Efficient Coflow Scheduling in Hybrid-Switched Data Center Networks

Efficient and Fair: Information-Agnostic Online Coflow Scheduling by Combining Limited Multiplexing with DRL

Fast Coflow Scheduling Via Traffic Compression and Stage Pipelining in Datacenter Networks

Online Scheduling of Coflows by Attention-Empowered Scalable Deep Reinforcement Learning.

Metaflow: A Better Traffic Abstraction for Distributed Applications.

A Case for Sampling Based Learning Techniques in Coflow Scheduling