In-Network Computing: A Survey of Recent Approaches and Challenges

In this article, the authors propose a new approach to congestion control in computer networks called In-Network Resource Pooling (INRP). The main idea is to pool resources within the network itself, rather than relying on external servers or cloud computing services. This approach can significantly improve network performance and reduce latency, making it ideal for applications that require real-time communication, such as virtual reality or autonomous vehicles.
The authors explain that traditional congestion control methods often rely on signaling between nodes in the network to adjust the amount of data being transmitted. However, these signals can be delayed or lost, leading to inefficiencies and increased latency. INRP addresses this issue by creating a set of mechanisms that ensure zero packet loss in intermediate network routers, maintain network stability, and increase link utilization.
The article provides several examples of how INRP can be applied in different scenarios. For instance, in mobile edge computing, INRP can be used to pool resources between nearby devices to improve data transfer efficiency. In cloud computing, INRP can help optimize resource allocation and reduce latency for applications that require real-time processing.
To illustrate how INRP works, the authors use a metaphor of a swimming pool. Imagine a group of people trying to share a limited amount of water in a pool. Without a coordinated system in place, some people may end up with too little water, while others have too much. INRP is like a lifeguard that ensures everyone gets an equal share of water, so the pool remains stable and efficient.
The authors also discuss the importance of collaboration between different stakeholders in the network to make INRP effective. They note that recent studies have focused mainly on edge computing, but ignoring the essential factor of resource management. INRP addresses this issue by providing a framework for managing 3C resources (computing, communication, and storage) across different nodes in the network.
In summary, INRP is a promising approach to congestion control that can significantly improve network performance and reduce latency. By pooling resources within the network itself, INRP can help optimize resource allocation and ensure efficient data transfer. The authors provide several examples of how INRP can be applied in different scenarios and emphasize the importance of collaboration between stakeholders to make it effective.