Latency-Aware Horizontal Computation Offloading for Parallel Processing in Fog-Enabled IoT

In this paper, we propose a two-step distributed horizontal architecture for computation offloading in a fogenabled Internet of Things (IoT) environment – HD-Fog – to minimize the overall energy consumption, and latency while executing hard real-time applications. The HD stands for the horizontal distribution of the tasks in the fog layer. Each sensor in the user devices independently captures data of varying formats. Parallel execution on these data is possible based on its Directed Acyclic Task Graph (DATG), and the corresponding results facilitate the ease of decision making. Towards this, in HDFog, the sensor nodes in user devices offload their tasks to a nearby fog node based on a greedy selection criterion. This fog node then further offloads the smaller sub-tasks, based on the DATG, among other fog nodes for parallel execution. Through extensive real-life metric-based emulation and comparison against traditional Fog and Cloud computing schemes, we observe that our approach 1) reduces the overall operational delays by 29% and 96%, and 2) offers promising speedup values. The proposed HD-Fog scheme also indicates a reduction in energy consumption by 30% compared to traditional fog computing schemes. Keywords—Computation Offloading, Fog Computing, Distributed and Parallel Computing, Queueing Theory, IoT