LTE HetNet Resource Management via Cell Dormancy and Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing resource management schemes in LTE heterogeneous networks (HetNets) are intractable due to non-ideal backhaul latency, which makes fast coordination among transmission points (TPs) impossible, and most methods either exploit cell dormancy or user association but not both efficiently.
Innovation Solution
A semi-static resource management approach that jointly exploits cell dormancy and load balancing by making some TPs inactive and associating users with active TPs at a coarse time-scale, using averaged metrics for coordination and fast-changing information for fine-time scale decisions, allowing for near-optimal user association and reduced interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fast coordination among TPs is implemented, then resource allocation efficiency is improved, but backhaul latency constraints make it impossible to achieve
Solution Approach 1:
The patent segments resource management into two distinct time scales: coarse time-scale (frame-level) for decisions requiring TP coordination, and fine time-scale (slot-level) for rapid adjustments. This segmentation allows critical coordination decisions to be made at a slower pace where backhaul latency is acceptable, while maintaining overall system efficiency through the finer-grained independent decisions at each TP.
Solution Approach 2:
The patent implements preliminary action by making user association decisions at the coarse time-scale before the actual data transmission occurs. User associations are determined in advance based on averaged metrics, allowing TPs to prepare their resource allocation strategies ahead of time, thus mitigating the impact of backhaul latency during the actual transmission phase.
2Loss of energy
If cell dormancy is exploited, then network energy efficiency is improved, but user association complexity increases
Solution Approach 1:
The patent applies dynamics by making the set of active TPs variable over time rather than fixed. TPs can transition between active and dormant states based on traffic conditions and user distribution, allowing the network to adaptively optimize energy efficiency while the user association algorithm dynamically adjusts to the changing active TP set.
Solution Approach 2:
The patent implements self-service by enabling each user to independently determine its optimal association with active TPs based on locally available channel state information and the published active TP set. This eliminates the need for complex centralized coordination for user association, reducing overall system complexity while still achieving energy-efficient cell dormancy exploitation.
3Object-affected harmful factors
If user association is optimized, then interference mitigation is improved, but computational complexity increases
Solution Approach 1:
The patent changes the parameter of time-scale granularity for user association decisions. By making associations at the frame-level rather than slot-level, the system reduces the frequency of optimization computations while still achieving effective interference mitigation. The averaged metrics over the frame period smooth out rapid fluctuations, reducing computational burden while maintaining performance.
4Adaptability or versatility
If fine-time scale user association is implemented, then resource allocation flexibility is improved, but coordination overhead increases
Solution Approach 1:
The patent segments the resource allocation flexibility into two layers: coarse-time-scale user association provides strategic flexibility in user-TP mapping, while fine-time-scale resource scheduling within each user association provides tactical flexibility for rapid channel variations. This segmentation reduces coordination overhead by limiting the scope of coordination to the coarser layer, while the finer layer operates independently at each TP.
Data Source
AI summary
At a coarse time-scale, at the start of each frame the choice of TPs to be made active and users to associate with the active TPs are determined by solving an optimization problem. The inputs to the optimization problem are averaged slowly varying metrics that are relevant for a period longer than the backhaul latency. At a the fine time-scale, in each slot each active TP independently does scheduling over the set of users associated with it, without any coordination with any of the other active TPs, based on fast changing information, such as instantaneous rate or SINR estimates.