Traffic congestion is a severe issue in urban areas, leading to long waiting times and increasing environmental pollution. To solve this issue, in the literature, authors proposed some traffic signal control systems using different techniques like Reinforcement Learning (RL), Deep Reinforcement Learning (DRL), and Federated Learning (FL). These techniques have their own pros and cons. In this study, we provide a comparative analysis of the existing techniques that are implemented to resolve traffic congestion. The comparison is done based on Average Waiting Time (AWT), Average Queue Length (AQL), computational complexity, training time, security, and generalizability parameters. Experimental results reveal that these techniques are capable of improving the traffic congestion problem, but have some drawbacks. At last, we provide the future scope for researchers to work on some shortcomings of the technique to develop a robust traffic signal control system.