Improving Fog Computing Infrastructure with Qos-Aware Resource Provisioning

In today's internet-enabled world, it's possible for everything to link to anything else. Because of this, a new way of working known as the Global Internet of Things has emerged, and it has begun to produce massive amounts of data. While it may be difficult to glean insights from streaming data, doing so has the potential to unleash technological marvels that will benefit society at large. It is critical to have a system in place to store, process, and forecast the data. Another active field of study, infrastructure development has seen a shift from centralised server systems to distributed cloud systems and is continuing to do so. There is a lot of room for investigation into the cloud's shortcomings in processing streaming data and making decisions in near real-time. Now that we have a network, the next step is to find a way to make it more useful and to distribute the load more evenly across the nodes via efficient routing. So that the network doesn't fail suddenly from too much traffic, the allocation is based on the present node load and the total network load. A balanced load on each node in the network is another result of optimised route selection. We built and tested a simple RL algorithm for an insignificant network that leverages the Q-Table. To get around the Q-Table's high level of complexity, an RL technique based on neural networks was suggested, which performs better on big networks. Over all time periods, the Proposed Deep RL technique outperforms Open Shortest Path First (OSPF) and Deep Neural Networks (DNN), achieving an accuracy of almost 92% in minimising average latency. In real-time applications, when minimising latency is crucial, its improved efficiency is evident.