Fog Computing characteristics have fascinated its services in many applicability areas such as Industrial IoT, health monitoring systems and high performance computing, weather forecasting, agronomy, etc. Here, data is captured using the various sensor nodes installed at different locations and further processed by a cloud-based server to obtain the desired results. Heterogeneity of fog nodes and end devices in fog computing architecture is a major feature that helps to utilize already existing computing resources, yet it complicates the managing process. Such a diversified environment must be considered when designing fog environments in terms of computing capabilities, power consumption, and connectivity. In this work An efficient Novel Particle Bee Optimization based Fault Tolerance framework has been projected that ensures the implementation of Fog computing in real time. The proposed framework handling live server faults and also diminishes latency and bandwidth usage of the network and improves Quality of Service (QoS). The achieved results for the projected technique represent the reduction of network bandwidth and reduction in overall cost.