Congestion on roads may lead to a catastrophe, especially for those large urban areas. Accordingly, different intelligent traffic-control methodologies had been implemented based on a variety of technologies such as DSP (Digital Signal Processing), WSNs (Wireless Sensors Networks), Image processing, etc. The design process depends on different factors such as fuel consumption, waiting time, traffic volume, and vehicle density. In this paper, we propose an adaptive traffic light control design based on hierarchal scheduling algorithm (WFQ/FCFS). The congestion control problem was modeled based on multi-agent systems, where the whole process was decomposed into a set of communicating sub agents. The traffic congestion control is based on minimizing the average total waiting time of vehicles at each lane in a single intersection. Our proposed control system is mainly designed and modeled based on packet switched networking model, where different classes of real-time traffics (video, audio) requesting the best QoS to be guaranteed. Simulation studies shows that the proposed adaptive-weighted-agent-based algorithm (AW Agent Based), the core of control design, outperforms the baseline algorithm as the variance in arrival rates increases.