Optimization For Network Connections

What is this patent about?

’634 is related to the field of network communication, specifically addressing the problem of slow bandwidth ramp-up in TCP/IP connections. Existing algorithms often take significant time to reach optimal bandwidth, leading to inefficiencies, especially in scenarios involving frequent connection setup and teardown, such as IoT devices, database access, and website content delivery. The patent aims to improve the speed at which full bandwidth is achieved for a given network connection.

The underlying idea behind ’634 is to leverage historical network performance data to intelligently initialize the transmission bandwidth for new connections. Instead of starting with a conservative bandwidth and gradually increasing it, the system analyzes past connections to similar endpoints or within the same geographical area and uses the performance data from those connections to set a more appropriate initial bandwidth. This reduces the time required to reach optimal throughput.

The claims of ’634 focus on a method, a non-transitory computer-readable medium, and a network-connected device that collect parameter values for determining bandwidth of network connections. A tuner server , distinct from the communicating endpoints, determines if a new connection matches a geographical area of a past connection. Based on this determination, the initial transmission bandwidth for the new connection is set using parameter values from the past connection.

In practice, the invention can be implemented in two primary ways: either the endpoint device itself stores and analyzes the historical network data, or a dedicated tuning server performs this function. In the endpoint-centric approach, the device compares the characteristics of the new connection with its stored data and sets the initial bandwidth accordingly. In the server-centric approach, the endpoint sends connection information to the tuning server, which then determines the appropriate bandwidth and sends it back to the endpoint.

The key differentiation from prior approaches lies in the proactive use of historical data to optimize initial bandwidth. Traditional TCP/IP implementations rely on reactive algorithms that adjust bandwidth based on real-time feedback, which inherently introduces a delay. By predicting the optimal bandwidth based on past performance, ’634 significantly reduces this ramp-up time, leading to improved communication efficiency, especially in scenarios with frequent short-lived connections. The use of a separate tuning server allows for a broader view of network conditions and potentially more accurate bandwidth predictions.