Traffic signal control is a critical and challenging problem, essential for minimizing urban traffic congestion and improving overall traffic flow. Current traffic signal control systems are limited by fixed timing schemes and inefficiencies. Recent advancements in deep reinforcement learning have shown promising potential for improving these systems. To better manage the complexities of urban traffic, multi-agent reinforcement learning (MARL) is well-suited for traffic signal control, enabling decentralized decision-making where each signal optimizes based on local conditions while coordinating with neighboring signals to reduce congestion. In this paper, we design and implement a MARL system with agents using three deep reinforcement learning algorithms, namely, Proximal Policy Optimization (PPO), Advantage Actor Critic (A2C), and Deep Q-Network (DQN), to control traffic signals in real time and compare the outcomes. Our results show that the PPO algorithm is the most effective, reducing average waiting times and increasing the average speed of vehicles in a traffic network.