A bandwidth control scheme for reducing the negative impact of bottlenecks in IoT environments: Simulation and performance evaluation

Abstract

The Internet of Things (IoT) environment comprises heterogeneous transmission channels. The statuses of these channels may change rapidly due to dynamic variations in things, data, topologies, etc. Thus, many bottlenecks may suddenly occur and dynamically change. Therefore, the quality of services (QoS) may be affected due to the lack of the bandwidth. Hence, this paper proposes a bandwidth control scheme to face the challenge of bottlenecks in the IoT environment. In this scheme, a bottleneck detection methodology, bandwidth prediction approach, reduction of bandwidth usage mechanism, and bandwidth management model are proposed. This bandwidth management model comprises reassigned and reallocated bandwidth plans. To test the proposed bandwidth control scheme, a large-scale simulation environment was constructed using NS-3. The performance of the proposed scheme was measured using the effect of reassigned and reallocated bandwidth plans in the cases of normal and prioritized data. In addition, packet loss, energy consumption, delay and bandwidth prediction accuracy were measured. Moreover, to make sure that the proposed scheme was positively effectiveness, its simulation results were compared to those of the famous machine learning and deep learning techniques: long short-term memory (LSTM), gated recurrent unit (GRU), autoregressive integrated moving average (ARIMA), multi-layer perceptron (MLP), and deep reinforcement learning (DRL). Finally, the simulation results proved that the proposed bandwidth control scheme notably outperformed the IoT environment's efficiency and limited the negative impact of the bottlenecks.