Distributed Media Classification via Local Controller Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The scalability of network systems that rely on a service controller for media classification is hindered by the overhead of transmitting multiple packets from local controllers, leading to increased processing demands as more devices and controllers are added.
Innovation Solution
Implementing a distributed system where local controllers perform media identification and generate a single metadata packet for each session, which is then transmitted to a service controller for classification, thereby separating media identification and classification logic and reducing the need for multiple packet transmissions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple packets are transmitted from local controllers to service controller for media classification, then media classification accuracy is improved, but system scalability deteriorates due to increased overhead
Solution Approach 1:
The patent combines multiple media packets from different local controllers into a single aggregated packet that contains classification-relevant information from all sources. This merging approach maintains classification accuracy by preserving essential data while reducing the number of transmissions to the service controller, thereby improving system scalability.
Solution Approach 2:
The invention extracts only the essential classification-related information from multiple media packets, separating it from redundant data. By taking out only the necessary features for media classification and transmitting them in a consolidated format, the system achieves accurate classification with reduced overhead, resolving the scalability contradiction.
2Adaptability or versatility
If centralized control is implemented in service controller, then service flexibility is improved, but processing overhead increases with more devices
Solution Approach 1:
The patent segments the control architecture by introducing intermediate aggregation points (local controllers) that perform preliminary processing and consolidation of media packets before forwarding to the service controller. This segmentation maintains centralized decision-making flexibility while distributing the processing workload, thereby reducing the energy overhead at the service controller as the network scales.
Data Source
AI summary
A network system and the method is described that splits media identification logic and media classification and/or general data session processing/management logic into separate controllers (e.g., between the local controllers and a service controller). By separating media identification logic and media classification logic (or general media processing/management logic), the system and method allows for greater flexibility and scalability in a network system employing a service controller. In particular, by allowing the local controllers to perform media identification and to construct a single packet that characterizes/describes a media session, the system avoids the need to transmit multiple data packets of a media session to the service controller for classification of the session. Accordingly, the system includes the flexibility a service controller provides while allowing for greater scalability by separating media identification logic and media classification respectively between a local controller and the service controller.