A Decentralized Latency-Aware Task Allocation and Group Formation Approach With Fault Tolerance for IoT Applications

Abstract

Development of internet of things (IoT) and smart devices eased life by offering numerous applications targeting to provide real-time low latency services, but they also brought challenges in handling huge data generated from the powerful computations, to get a job done. Decentralized edge computing could help to achieve latency requirements of the applications by executing them closer to the user at edge of network, but most of the current studies actually deployed centralized approaches for cluster computing at edge, which put extra overhead of cluster formation and management. In this article, we propose to group heterogeneous edge nodes on task arrival with a more decentralized method and execute tasks in parallel to meet their deadline. On the other hand, to guarantee successful execution of critical IoT application running in an edge network, fault tolerance has to be significantly considered. For resource limited edge devices, there is a great need for efficient fault tolerance techniques, which can provide reliability based on device's local information, without worrying about overall network topology. In this article, our novel method is to increase task reliability being executed in distributed edge computing environment through finding reliability of an edge node locally, and by providing fault tolerance to increase overall application availability. Our proposed fault tolerance technique works in decentralized mode by executing new algorithms to handle above mentioned problems. Our experiment results show that our approach is effective as well as providing desired goals of achieving deadline for latency-aware IoT applications, with staggering decrease in overall network traffic along with ensuring reliability and availability.

Fault Tolerance in IoT: Techniques and Comparative Study

“…In passive replication there is only one server (called primary) that processes client requests. In paper [12], fault tolerance is achieved by dividing end nodes into different groups, within each group all other nodes act as a backup node for each node.…”

“…How a system is going to perform in different scenarios by considering some set of measures like time constraints, environment, etc. In paper [12], distributed edge computational network is being discussed in which, by introducing preprocessing at end nodes, much of the computation is handled at the end nodes itself, hence reducing network traffic in between end nodes and the central server. This also helped in reducing network utilization and decreasing network latency.…”

“…In papers [4], [8] active replication is being employed whereas in paper [10], [11], [12] passive replication is being employed.…”

“…When the traffic volume is greater than what can be transmitted over the network, devices queue, or keep, the packets are held in memory before the resources are made available to transmit them. In papers [8], [9], [11], [12] performance is focused. Availability is used as an attribute in papers [8], [9], [10], security is described in papers [1], [4], scalability is used as an attribute in paper [9], interoperability is used as an attribute in paper [4], energy consumption is focussed in paper [9].…”

A Quick Algorithm to Determine 2-Optimality Consensus for Collectives

“…IoT has been rapidly developed during recent years. Its applications have the potential to affect every aspect of daily human life [42], many consensus problems exist in this field, for example, resource allocation [43], task allocation [44], [45], decision making in service-oriented [46].…”