A literature review on designing and managing network congestion control must consider various factors, including theoretical foundations, practical methods, and modern challenges. Initial research concentrated on reactive methods, such as TCP algorithms like Tahoe, Reno, and New Reno, which respond to packet loss by adjusting the sending rate. These methods have limitations, such as slow reactions to congestion, unfairness among data streams, and difficulties in high-speed, low-delay environments.
Then came Active Queue Management (AQM) methods, such as Random Early Detection (RED). These aimed to manage the queue lengths at routers to prevent packet loss and notify end systems of potential congestion, improving overall network performance. Yet, RED has its downsides, like needing careful parameter tuning and potentially favoring certain types of traffic, which pushed researchers to find more advanced AQM techniques.
There is increasing interest in using model-based and machine-learning techniques for network congestion control. Model-based methods use mathematical models to predict and manage congestion, but creating accurate models can be difficult. Machine learning, especially reinforcement learning, analyzes historical data to adapt to traffic and resource management. However, it has challenges like requiring large data sets and understanding models.
The literature highlights the challenge of balancing fairness and efficiency in network congestion control design. Traditional algorithms often prioritize fairness, risking inefficiency, while focusing on efficiency can result in unfair bandwidth distribution. Recent studies explore methods to achieve both goals.
Finally, the rise of Software-Defined Networking (SDN) and Network Function Virtualization (NFV) has given new ways to manage network congestion. This allows for centralized control and the ability to program network resources, which helps in creating smarter and more flexible network congestion control methods. Ongoing research in this area is vital for keeping networks running smoothly as they grow more complex.
Click here to get the complete project:
