English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In edge computing, users' service profiles must be migrated in response touser mobility. Reinforcement learning (RL) frameworks have been proposed to doso. Nevertheless, these frameworks do not consider occasional server failures,which although rare, can prevent the smooth and safe functioning of edgecomputing users' latency sensitive applications such as autonomous driving andreal-time obstacle detection, because users' computing jobs can no longer becompleted. As these failures occur at a low probability, it is difficult for RLalgorithms, which are inherently data-driven, to learn an optimal servicemigration solution for both the typical and rare event scenarios. Therefore, weintroduce a rare events adaptive resilience framework FIRE, which integratesimportance sampling into reinforcement learning to place backup services. Wesample rare events at a rate proportional to their contribution to the valuefunction, to learn an optimal policy. Our framework balances service migrationtrade-offs between delay and migration costs, with the costs of failure and thecosts of backup placement and migration. We propose an importance samplingbased Q-learning algorithm, and prove its boundedness and convergence tooptimality. Following which we propose novel eligibility traces, linearfunction approximation and deep Q-learning versions of our algorithm to ensureit scales to real-world scenarios. We extend our framework to cater to userswith different risk tolerances towards failure. Finally, we use trace drivenexperiments to show that our algorithm gives cost reductions in the event offailures.

FIRE: A Failure-Adaptive Reinforcement Learning Framework for Edge Computing Migrations