Role of Middleware in Improving Reliability of Fog Applications

Abstract

Fog computing extends cloud services to the network edge to address latency and reliability challenges in modern applications. However, resource management in fog environments remains a challenge due to heterogeneity and the dynamic nature of fog nodes. While middleware has traditionally addressed similar challenges in distributed computing, its application to fog computing resource management has not been thoroughly explored. Therefore, this thesis proposes a middleware approach to resource management in fog computing.

A novel adaptive middleware architecture employing the MAPE-K(Monitor, Analyse, Plan, Execute over shared Knowledge) self-adaptation framework to dynamically switch between clustered and peer-to-peer configurations based on conditions in the environment. The middleware introduces: (1) a request handling algorithm that implements multi-tier resource discovery across fog nodes, clusters, and cloud layers with O(1) to O(S) complexity depending on service lookup implementation; (2) an Autonomous System-inspired clustering mechanism with dedicated inter-cluster communication interfaces for cross-domain resource sharing; and (3) a decentralised P2P bootstrapping algorithm for optimal peer selection based on propagation delay and computational capacity.

The proposed solutions are evaluated through simulations using iFogSim and Omnet++. Simulation experiments include processing over 3,500 IoT requests following an exponential distribution with a mean inter-arrival time, λ = 100ms and 3-20 fog nodes to evaluate the performance of centralised and decentralised architectures. The simulation results demonstrate improvements in response time, network utilisation, and energy efficiency compared to traditional cloud-centric and static/non-collaborative fog architectures. The results highlight the potential of middleware in improving adaptability and reliability of fog computing.

This work’s key contributions are: (1) First middleware framework enabling both
clustered and P2P fog architectures with adaptive switching; (2) Novel inter-
cluster resource sharing mechanisms addressing the gap in fog clustering research; (3) Comprehensive resource management framework integrating task offloading, load balancing, and service discovery.

Date of Award	19 Jun 2025
Original language	English
Awarding Institution	Edge Hill University
Supervisor	ELLA PEREIRA (Director of Studies), MOHSIN RAZA (Supervisor) & UMAR KHAN (Supervisor)