Distributed Local Caching for Access Network Bandwidth Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for bandwidth due to large amounts of electronic content data transfer from popular websites, such as YouTube, causes congestion and latency in access networks, and existing solutions like caching do not scale well with the number of concurrent requests, often shifting the I/O bottleneck from servers to caching nodes.
Innovation Solution
A method where the local storage of end-user devices is populated with requested electronic content information, and an access network registry monitors data traffic to redirect requests from one end-user device to another that has already stored the content, thereby distributing the I/O bandwidth among end-users and reducing the load on content providers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If content is cached at centralized nodes in the access network, then bandwidth requirements are reduced, but device complexity and I/O bottleneck shift to caching nodes
Solution Approach 1:
The patent extracts the caching function from centralized access network nodes and distributes it to individual end-user devices. Each device maintains its own local cache, eliminating the need for complex centralized caching infrastructure while reducing bandwidth consumption through local content reuse.
Solution Approach 2:
End-user devices autonomously manage their own caching without requiring centralized control or coordination. Each device independently determines what content to cache and serves content from its local storage, eliminating the I/O bottleneck that would otherwise occur at centralized caching nodes.
2Reliability
If unicast protocol is used for content delivery, then each client receives content from server, but scalability deteriorates with large numbers of concurrent requests
Solution Approach 1:
The patent merges multiple unicast streams into a single source by having end-user devices share content among themselves. When one device receives content via unicast from the server, it becomes a source for other devices, effectively combining multiple delivery paths and enabling scalable content distribution.
Solution Approach 2:
Content is pre-fetched and stored in local caches at end-user devices before being requested by others. This preliminary action allows devices to serve content locally without requiring real-time server communication, dramatically improving scalability for concurrent requests.
3Loss of energy
If distributed caching is implemented at end-user devices, then bandwidth requirements are reduced, but loss of information increases due to cache invalidation challenges
Solution Approach 1:
The patent implements feedback mechanisms where devices report cache status and content availability to the network. This feedback enables the system to track distributed cache contents, manage invalidation appropriately, and ensure content availability while maintaining bandwidth efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A source on a data network supplies an item of electronic content information via an access network to a plurality of data processing devices. Each data processing device has a local storage. A node on the access network comprises a registry. The node monitors a request issued by a specific data processing device to the source for delivery of the item. The node consults the registry for determining whether the item is available from another data processing device. If the item is available from the other data processing device, the node redirects the request to the other data processing device, and registers in the registry an availability of the item from the specific data processing device when the other data processing device has supplied the requested item. If the item is not available from any data processing device, the node forwards the request to the source, monitors a response from the source, and registers in the registry the availability of the item from the specific data processing device when the source has delivered the item.