Intelligent Application Learning for Data Bundling and Fast Dormancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
UMTS networks face significant growth in data signaling load due to frequent state transitions and independent data connections from user equipment (UE), leading to increased RNC processing load and battery consumption, despite efforts to conserve resources through fast dormancy and inactivity timers.
Innovation Solution
Implementing an intelligent application learning mechanism that profiles user equipment applications to predict data flow arrival times and bundle data sessions, delaying fast dormancy transitions when subsequent data flows are anticipated, thereby reducing the number of radio resource control connections and signaling events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If fast dormancy is implemented to quickly transition UE from DCH to IDLE state, then battery life is conserved, but RNC processing load increases due to frequent connection reestablishment
Solution Approach 1:
The system performs preliminary learning and profiling of application behavior patterns before making state transition decisions. By analyzing historical data and predicting future data flow arrivals, the system prepares in advance to determine whether fast dormancy should be applied, avoiding premature state transitions that would require costly reestablishment
Solution Approach 2:
The system implements a feedback mechanism where application behavior is continuously monitored and learned, and this information feeds back into the state transition decision-making process. The RNC uses learned application profiles to adaptively control fast dormancy, creating a closed-loop system that optimizes both battery life and processing load over time
2Adaptability or versatility
If multiple independent data connections are established for different applications, then each application's data requirements are met, but signaling load and RNC processing requirements increase
Solution Approach 1:
The system merges multiple independent data connections into a single bundled connection when applications exhibit compatible behavior patterns. By learning that certain applications frequently access data simultaneously or sequentially, the system combines their signaling into one unified connection, reducing overall signaling load while maintaining data delivery for all applications
Solution Approach 2:
The learned application profiles serve multiple functions: they predict data flow arrivals for bundling decisions, determine fast dormancy applicability, and optimize resource allocation. This universal profiling mechanism handles diverse application types (streaming, browsing, messaging) through a single adaptive framework
3Productivity
If inactivity timers are set to short values for resource efficiency, then resources are conserved, but user perceived latency increases
Solution Approach 1:
The system performs preliminary prediction of data flow arrivals using learned application patterns before the inactivity timer expires. By anticipating when data will arrive, the system can extend the timer only when necessary, maintaining resource efficiency while preventing premature disconnection that would cause latency
Solution Approach 2:
The inactivity timer value becomes dynamic rather than static, adjusting based on learned application behavior. For applications with predictable periodic data flows, the timer is extended to match the expected interval. For applications with uncertain patterns, the timer remains short to conserve resources, creating a dynamic adaptation mechanism
Data Source
AI summary
Data bundling and fast dormancy controls are provided based on application monitoring and classification. Moreover, a balance is enabled between saving battery power of a user equipment (UE) and reducing signaling and processing load in a radio resource controller (RRC). For instance, a system can observe data flow related behavior of applications on the UE. On receiving a first data flow request, an arrival time of a next data flow request is predicted based on an analysis of the behavior, and the system determines whether the two data flows can be bundled together and transmitted over a single connection. Additionally, on completion of the first data flow, the arrival time of the next data flow request is predicted based on the analysis, and the system determines whether a fast dormancy timer can be disabled to transmit the next data flow over the current connection.