Evaluation of Machine Learnable Bandwidth Allocation Strategy for User Cooperative Traffic Forwarding

Abstract

In recent years, user cooperative traffic forwarding is a popular study topic and broadly seen as one of the important promising technologies to improve energy efficiency (EE) of the battery-driven mobile terminal (MT). However, the battery-driven devices always suffer from a problem of limited working time due to battery life. In this paper, we propose a simply machine learnable bandwidth allocation strategy for user cooperation-aided wireless communication systems and evaluate the power consumption of the systems via both theoretical and experimental approaches. By using the proposed bandwidth allocation strategy, we first derive the mathematical expressions to evaluate the transmission power of the MTs for non-cooperative and cooperative scenarios by a generalized channel model. In this generalized model, the spatially correlated shadowing and frequency selective fading are considered as channel effects, and this generalized model is mathematically analyzed for the consumed power via the proposed scenarios with the long-term evolution (LTE) power model for smartphones. In the final stage, we evaluate the results by our smartphone test-bed. The results obtained in this paper show that the benefits of the user cooperation-aided traffic forwarding are significant. Unfortunately, according to the numerical analysis, because there are some physical constraints for MTs, such as maximal transmit power, we cannot drastically obtain the benefits in real application cases. Some interesting points, such as how to use a machine learning approach to reduce the system complexity and thus improve transmission performances, are also discussed in this paper.

Keywords

• User cooperation
• machine learning
• device-to-device communication
• frequency selective fading
• power consumption
• energy efficiency