Network Utility Aware Traffic Loading Balancing in Backhaul-constrained Cache-enabled Small Cell Networks with Hybrid Power Supplies

Tao Han, Nirwan AnsariarXiv 2014 · arXiv

TL;DR

This paper proposes a novel "Network Utility Aware" (NUA) traffic load balancing scheme for small cell networks that are constrained by backhaul capacity and powered by hybrid energy sources. The NUA scheme dynamically optimizes user association to balance green power utilization and traffic delivery latency, considering factors like BS capacity, backhaul constraints, and cache performance.

What Problem Does This Paper Address?

The rapid growth of mobile data traffic necessitates increased network capacity, leading to the widespread deployment of small cell base stations (SCBSs). However, SCBSs often face backhaul constraints, meaning their connection to the core network is a bottleneck. Additionally, the dense deployment of SCBSs can lead to significant energy consumption. To address these issues, researchers are exploring techniques like content caching within SCBSs to reduce backhaul load and the use of renewable energy sources to power these stations.

The challenge lies in effectively managing traffic load in such complex networks. Simply optimizing for one factor (like green energy usage) can negatively impact others (like latency). This paper tackles the problem of dynamically balancing traffic load in backhaul-constrained, cache-enabled small cell networks that utilize hybrid power supplies (combining grid power and renewable energy). The goal is to find an optimal balance between green power utilization and overall network performance, specifically traffic delivery latency.

Methods

The core of this paper is the proposed Network Utility Aware (NUA) traffic load balancing scheme. This scheme operates through a distributed, four-phase procedure:

  1. Initial User Association and Network Utility Measurements: Users connect to a BS to gather network utility information. BSs then measure their traffic load and backhaul load.
  2. Network Utility Advertisement: BSs advertise their measured network utility information.
  3. User Selection: Users select their serving BS based on advertised network utility information and downlink data rates, aiming to minimize a defined objective function.
  4. Iterative Updates: BSs and users iteratively update their network utilities and BS selections until the user association converges.

The NUA scheme considers four key network utilities:

  • Green Power Utilization: Minimizing brown power consumption.
  • Traffic Delivery Latency in BSs: The time it takes for data to be processed within a BS.
  • Traffic Delivery Latency in Backhaul: The time it takes for data to travel through the BS's backhaul.
  • Cache Hit Ratio: The effectiveness of the BS's content caching system.

The scheme formulates a traffic load balancing problem as an optimization problem aiming to minimize a weighted sum of these utilities. A key aspect is the introduction of a system parameter, $\kappa$, which allows for an adjustable tradeoff between green power utilization and traffic delivery latency.

Datasets

Not discussed in the paper. The paper relies on simulations using a defined network model and parameters.

Key Insights and Ideas

  • The Interplay of Utilities is Crucial: The paper highlights that optimizing individual network utilities (like green power or latency) in isolation can lead to suboptimal overall performance. A holistic approach considering multiple utilities is essential.
  • Backhaul and Cache are Key Bottlenecks: For small cell networks, the backhaul capacity and the effectiveness of content caching are critical factors that significantly influence traffic load balancing and overall network performance.
  • Hybrid Power Management is Complex: Balancing the use of intermittent renewable energy with reliable grid power requires intelligent load management to minimize brown power consumption without sacrificing network performance.
  • Dynamic Tradeoffs are Necessary: The NUA scheme's ability to dynamically adjust the tradeoff between green power utilization and traffic delivery latency based on network conditions (controlled by $\kappa$) is a significant advantage.
  • User Association is the Lever: The paper demonstrates that intelligently guiding user association is the primary mechanism for achieving the desired traffic load balancing and utility optimization.

Key Results

  • The proposed NUA scheme demonstrates convergence to an optimal user association within a reasonable number of iterations (less than 100).
  • The NUA scheme significantly reduces traffic delivery latency, especially when the network is constrained by backhaul data rates.
  • By adjusting the system parameter $\kappa$, the NUA scheme can effectively control the tradeoff between brown power consumption and traffic delivery latency. A higher $\kappa$ prioritizes green power utilization, leading to less brown power consumption but potentially higher latency.
  • The NUA scheme outperforms the DRB-NU (Data Rate Bias - Network Utility awareness) scheme in terms of reducing traffic delivery latency and brown power consumption. For example, it reduced latency by 8.7% and brown power consumption by 20.28%.
  • Cache awareness in the NUA scheme significantly helps reduce traffic delivery latency, particularly when backhaul data rates are low.

Limitations and Future Work

  • Residual Green Power: The paper states that the redistribution of residual green power is not considered in the current model, which is out of the scope of this paper.
  • Cache Hit Ratio Estimation: The paper assumes that the cache hit ratio can be estimated for the duration of a user association process. The actual optimization and estimation of cache hit ratios are considered out of scope.
  • Time Scale of Traffic Load Balancing: The assumption that traffic load balancing occurs at a time scale of several minutes, while solar power generation is modeled at an hourly scale, simplifies the analysis but might not fully capture rapid fluctuations in renewable energy availability.

Future work could explore:

  • Dynamic cache management strategies to further optimize performance.
  • More sophisticated models for renewable energy generation and consumption.
  • Investigating the impact of mobility and user behavior on the proposed scheme.
  • Extending the model to consider other network utilities or constraints.

Summary updated

BibTeX
@misc{han2014network,
  author = {Tao Han and Nirwan Ansari},
  title = {{Network Utility Aware Traffic Loading Balancing in Backhaul-constrained Cache-enabled Small Cell Networks with Hybrid Power Supplies}},
  year = {2014},
  eprint = {1409.8267},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1409.8267},
}
Figure