Path Segmentation-Based Hybrid Caching in Information-Centric Networks

Wěi Li, Peng Sun, Rui HanFuture 2021 · DOI

TL;DR

This paper introduces a novel hybrid caching strategy for Information-Centric Networks (ICNs) called Path Segmentation-Based Hybrid Caching (PSBC). PSBC dynamically divides data return paths into "user-edge" and "non-edge" areas, applying different caching strategies to each to simultaneously reduce user access latency and improve network resource utilization.

What Problem Does This Paper Address?

Traditional host-centric networks struggle with the increasing demand for content distribution and mobility. Information-Centric Networks (ICNs) offer a promising alternative with features like in-network caching. However, effectively managing and placing cached content is crucial due to limited cache space and massive internet traffic. Existing caching schemes often focus on either minimizing user access latency or maximizing network resource utilization, but rarely achieve both optimally. This paper aims to bridge this gap by proposing a hybrid approach that balances these two critical objectives.

Methods

The core innovation of PSBC lies in its path segmentation strategy. Instead of treating all nodes on a data return path equally, PSBC dynamically divides the path into two segments:

  • User-Edge Area: This segment is closer to the end-user. Here, the focus is on minimizing user access latency. A heuristic caching scheme is employed, prioritizing popular content to ensure quick retrieval. Cache decisions in this area are made individually by each node, considering content popularity and transmission distance.
  • Non-Edge Area: This segment is further from the user. The primary goal here is to improve network resource utilization. This area acts as a secondary cache for the user-edge area. When content is replaced in the user-edge area due to space constraints, it's migrated to a selected node in the non-edge area. This migration is managed through a "path competition" mechanism where nodes in the non-edge area collaboratively select the most suitable migration destination based on their "cache pressure value."

The paper also introduces a cache value (CV) calculation for content, which considers both data transmission distance and content popularity. This CV guides caching and replacement decisions in the user-edge area. For migrated content in the non-edge area, a simpler popularity-based CV is used.

Datasets

Not discussed in the paper. The evaluation uses simulated network topologies.

Key Insights and Ideas

  • Dynamic Path Segmentation is Key: The paper highlights that a one-size-fits-all caching approach is suboptimal. Dynamically segmenting the data return path based on proximity to the user allows for tailored caching strategies that address different network objectives.
  • User-Edge for Latency, Non-Edge for Efficiency: The clear division of roles between the user-edge (low latency) and non-edge (resource utilization) areas is a smart way to balance competing goals.
  • Content Migration as a Resource Optimizer: Migrating less popular or less frequently accessed content from the edge to the non-edge area frees up valuable cache space at the edge for more popular items, improving overall cache efficiency.
  • Cache Pressure Value for Smart Migration: The "cache pressure value" metric for nodes in the non-edge area provides a practical way to select the best migration destination, avoiding the need for extensive communication overhead.

Key Results

The simulation results demonstrate that PSBC outperforms other popular caching strategies (LCE, LCD, ProbCache, CL4M) in terms of both cache hit ratio and average access latency.

  • Cache Hit Ratio: PSBC consistently achieves a higher cache hit ratio across different network topologies and skewness values (representing content popularity distribution). For instance, with a skewness of 0.8, PSBC's cache hit ratio was significantly higher (e.g., 48% higher than LCE in the GEANT topology). This is attributed to PSBC's focus on popular content in the user-edge area and efficient resource utilization.
  • Average Access Latency: PSBC also shows lower average access latency. This is because nodes closer to users are optimized to cache popular content, directly reducing the time it takes for users to retrieve their desired information. The paper shows PSBC's latency is lower than other strategies across various cache size ratios.

Limitations and Future Work

The authors acknowledge that the current path segmentation proportion is set with a fixed parameter (k=2) for convenience and suggests that optimizing this proportion could be a valuable future research direction. They also plan to explore a more flexible path segmentation algorithm and test the PSBC scheme in other application scenarios. The current work is based on a preliminary simulation, and further real-world testing would be beneficial.

Summary updated

Abstract

Information-centric networks (ICNs) have received wide interest from researchers, and in-network caching is an important characteristic of ICN. The management and placement of contents are essential due to cache nodes’ limited cache space and the huge Internet traffic. This paper focuses on coordinating two cache metrics, namely user access latency and network resource utilization, and proposes a hybrid caching scheme called the path segmentation-based hybrid caching scheme (PSBC). We temporarily divide each data transmit path into a user-edge area and non-edge area. The user-edge area adopts a heuristic caching scheme to reduce user access latency. In contrast, the non-edge area implements caching network content migration and optimization to improve network resource utilization. The simulation results show that the proposed method positively affects both the cache hit ratio and access latency.
BibTeX
@inproceedings{li2021path,
  author = {Wěi Li and Peng Sun and Rui Han},
  title = {{Path Segmentation-Based Hybrid Caching in Information-Centric Networks}},
  booktitle = {Future},
  year = {2021},
  doi = {10.3390/fi13010016},
}
Figure