Dynamic Application Placement Technique for Intelligent Heterogeneous Fog Environment

Fog computing has gained immense popularity as it brings computing and processing close to the user’s devices and harnesses the system’s capability. However, the fog nodes are heterogeneous with different capacities and capabilities, resulting in many challenges including optimal placement of the application modules over the fog nodes and effectively mapping the nodes with the required resources for the tasks. Therefore to tackle these challenges, this paper proposes a technique for application module placement in a fog environment with heterogeneous nodes while optimizing bandwidth consumption and resource utilization. The proposed application placement technique uses deep Q-learning to predict the pattern of incoming requests and Markov Chain is used for modelling. OpenAI Gym framework is used for the simulation. Low-intensive and time-sensitive applications are executed on the appropriate fog nodes that are close to user’s device requesting the service. From the simulation results, it is observed that proposed technique outperforms the existing algorithms such as First Come First Serve and Fuzzy based algorithms, considering the fog environment’s dynamicity, mobility, and heterogeneity characteristics.