Abstract
We study the problem of balancing timeliness and criticality when gathering data from multiple sources using a two-level hierarchical approach. The devices that generate the data transmit them to a local hub. A central decision maker then has to decide which local hubs to allocate bandwidth to and the local hubs have to prioritize the messages they transmit when given the opportunity to do so. Whereas an optimal policy does exist for this problem such a policy would require global knowledge of messages at each local hub, rendering such a scheme impractical. We propose a distributed reinforcement-learning-based approach that accounts for both the timeliness requirements and criticality of messages. We evaluate our solution using a criticality-weighted deadline miss ratio as the performance metric. The performance analysis is done by simulating the behavior of the proposed policy as well as that of several natural policies under a wide range of system conditions. The results show that the proposed policy outperforms all the other policies - except for the optimal but impractical policy - under the range of system conditions studied and that in many cases it performs close (3% to 12% lower performance depending on the condition) to the optimal policy.
Highlights
With the proliferation of devices that can gather and transmit data about our world, we are confronted with the challenge of collecting and processing this data
In the system architecture we consider, data is collected in a two-level setup: devices may communicate with a local hub, and local hubs communicate with a global entity
Local hubs collect messages that need to be transmitted to the central hub, and need to decide on how to prioritize these messages for transmission to the central hub, which is at the top of this hierarchical system
Summary
With the proliferation of devices that can gather and transmit data about our world, we are confronted with the challenge of collecting and processing this data Many such devices that we may deploy to observe and control aspects of our physical environment will use wireless communication links, and the available bandwidth for data transmissions can be saturated. We propose and evaluate a scheduling mechanism for a two-level network architecture where messages that need to be transmitted have an associated deadline and criticality level. Each hub maintains a finite number of message queues, and messages can be assigned to any of these queues These queues have fixed priorities associated with them, and data is transmitted to the global entity according to this prioritization; each hub will empty a high priority queue first before moving to a queue with lower priority data. Propose a distributed decision making policy that is based on reinforcement learning
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have