Quality and Cost Aware Service Selection in IoT-Context Management Platforms
Kanaka Sai Jagarlamudi, Arkady Zaslavsky, Seng W. Loke, Alireza Hassani, Alexey Medvedev · 2021
The IoT-context management platforms (CMPs) have evolved from IoT silos to advanced systems that seamlessly access context providers (third-party IoT-data providers). Such evolution enabled them to respond adequately to dynamically varying needs of context consumers (context-aware IoT appli-cations). Nevertheless, the densely deployed context providers lead to multiple options that potentially complete context re-quests, resulting in the problem of 'which context provider to select ?‘. Moreover, these providers often vary in their service characteristics-Cost of Context (CoC) and Quality of Context (QoC). Therefore, the CMPs should be enabled with QoC and CoC-aware selection mechanisms to select effective context providers. Unfortunately, the Service Level Agreements (SLAs) between the context providers and CMPs that helps design these mechanisms have substantial shortcomings. This paper proposes an SLA template along with a CoC-aware selection mechanism called the Assurance. Using this SLA template, the context providers negotiate their QoC and CoC levels with the CMP. Hence, by having SLAs as a base, the CMP developers can design effective selection mechanisms. When implemented with QoC-aware selection mechanisms, Assurance improves the CMP's cost-effectiveness by selecting the low-cost context providers that deliver the required QoC. This paper also proposes metrics to measure the QoC and CoC outcomes from the invoked provider at the CMP level. Furthermore, it presents a framework that implements these contributions and performs QoC and CoC-aware selection. The evaluation demonstrates that our selection framework improves QoC outcomes and cost-effectiveness com-pared to using a naive approach and QoC-aware selection models without Assurance.