[Paper Review] Throughput-Delay Tradeoffs in Content-Centric Wireless Networks

Milad Mahdian, Edmund Yeh|arXiv (Cornell University)|Apr 15, 2015

Caching and Content Delivery12 references3 citations

TL;DR

This paper analyzes throughput-delay tradeoffs in content-centric wireless networks with cache-enabled nodes and base stations, proposing an optimized caching scheme under Zipf popularity. It shows that heterogeneous networks outperform homogeneous ones only when base station density exceeds the node-to-content ratio, and that performance gains vanish if the Zipf exponent exceeds 1.5.

ABSTRACT

We study the throughput and delay characteristics of wireless networks based on a content-centric network architecture, where users are mainly interested in retrieving content stored in the network, rather than in maintaining source-destination communication. Nodes are assumed to be uniformly distributed in the network area. Each node has a limited-capacity content store, which it uses to cache contents according to the proposed caching scheme. Requested content follows a general popularity distribution, and users employ multi-hop communication to retrieve the requested content from the closest cache. We derive the throughput-delay tradeoff of the content-centric wireless network model and solve the caching optimization problem. We then evaluate the network performance for a Zipf content popularity distribution, letting the number of content types and the network size both go to infinity. Finally, we extend our analysis to heterogeneous wireless networks where, in addition to wireless nodes, there are a number of base stations uniformly distributed at random in the network area. We show that in order to achieve a better performance in a heterogeneous network in the order sense, the number of base stations needs to be greater than the ratio of the number of nodes to the number of content types. Furthermore, we show that the heterogeneous network does not yield performance advantages in the order sense if the Zipf content popularity distribution exponent exceeds 3/2.

Motivation & Objective

• To understand the fundamental throughput-delay tradeoffs in content-centric wireless networks where users retrieve cached content instead of establishing end-to-end connections.
• To optimize content caching at nodes under general content popularity distributions, particularly Zipf distributions.
• To evaluate the performance gain of introducing base stations in heterogeneous wireless networks compared to purely ad hoc content-centric networks.
• To determine the scaling conditions under which heterogeneous networks achieve better order-of-magnitude performance than homogeneous ones.

Proposed method

• Models a wireless network with uniformly distributed nodes, each equipped with a limited-capacity cache, using a multi-hop routing scheme to retrieve content from the nearest cached copy.
• Applies a caching policy that prioritizes popular content based on a general popularity distribution, with analytical derivation of throughput and delay under this scheme.
• Uses asymptotic analysis as both the number of nodes and content types grow to infinity, focusing on the Zipf distribution for content popularity.
• Extends the model to include randomly distributed base stations, analyzing how their density affects network performance in the order sense.
• Derives conditions under which the heterogeneous network achieves superior performance scaling, based on the ratio of nodes to content types and the Zipf exponent.

Experimental results

Research questions

• RQ1 What is the fundamental throughput-delay tradeoff in a content-centric wireless network with cache-enabled nodes and multi-hop content retrieval?
• RQ2 How does the caching policy affect network performance under a general content popularity distribution, particularly Zipf?
• RQ3 Under what conditions does the inclusion of base stations in a heterogeneous network improve performance in the order sense compared to a purely ad hoc network?
• RQ4 How does the Zipf exponent of content popularity influence the performance scaling of heterogeneous content-centric networks?

Key findings

• The heterogeneous network achieves better performance in the order sense only when the number of base stations exceeds the ratio of the number of nodes to the number of content types.
• If the Zipf exponent of content popularity exceeds 3/2, the heterogeneous network does not yield any performance advantage in the order sense over a homogeneous network.
• Throughput and delay are shown to scale with network size and content popularity distribution, with explicit asymptotic expressions derived as the number of nodes and content types grow large.
• The caching optimization problem is solved under the given model, showing that content popularity and cache capacity jointly determine the optimal caching strategy.
• The multi-hop retrieval mechanism from the nearest cached copy significantly reduces delay compared to long-haul source-destination communication.
• Performance gains from base stations are strictly bounded by the scaling of node and content densities, indicating a fundamental limit to infrastructure benefits.