Service Provisioning in Mobile Environments through Opportunistic Computing

TL;DR

This paper introduces an analytical model-based opportunistic computing algorithm that efficiently ranks and selects services in mobile networks, significantly reducing service provisioning times across various dynamic scenarios.

Contribution

It presents a novel service ranking algorithm based on an analytical model for opportunistic computing in mobile environments, improving service selection efficiency.

Findings

The algorithm accurately ranks services based on expected completion time.

It achieves lower service provisioning times compared to other policies.

Performance remains robust across different mobility and resource conditions.

Abstract

Opportunistic computing is a paradigm for completely self-organised pervasive networks. Instead of relying only on fixed infrastructures as the cloud, users' devices act as service providers for each other. They use pairwise contacts to collect information about services provided and amount of time to provide them by the encountered nodes. At each node, upon generation of a service request, this information is used to choose the most efficient service, or composition of services, that satisfy that request, based on local knowledge. Opportunistic computing can be exploited in several scenarios, including mobile social networks, IoT and Internet 4.0. In this paper we propose an opportunistic computing algorithm based on an analytical model, which ranks the available (composition of) services, based on their expected completion time. Through the model, a service requesters picks the one that is expected to be the best. Experiments show that the algorithm is accurate in ranking services, thus providing an effective service-selection policy. Such a policy achieves significantly lower service provisioning times compared to other reference policies. Its performance is tested in a wide range of scenarios varying the nodes mobility, the size of input/output parameters, the level of resource congestion, the computational complexity of service executions.