Renewable Energy-Aware Information-Centric Networking

The ICT industry today is placed as one of the major consumers of energy, where recent reports have also shown that the industry is a major contributor to global carbon emissions. While renewable energy-aware data centers have been proposed, these solutions have certain limitations. The primary limitation is due to the design of data centers which focus on large-size facilities located in selected locations. This paper addresses this problem, by utilizing in-network caching with each router having storage and being powered by renewable energy sources (wind and solar). Besides placing contents closer to end users, utilizing in-network caching could potentially increase probability of capturing renewable energy in diverse geographical locations. Our proposed solution is dual- layered: on the first layer a distributed gradient-based routing protocol is used to discover the paths along routers that are powered by the highest renewable energy, and on the second layer, a caching mechanism will pull the contents from the data centre and place them on routers of the paths that are discovered by our routing protocol. Through our experiments on a testbed utilizing real meteorological data, our proposed solution has demonstrated increased quantity of renewable energy consumption, while reducing the workload on the data centers.

💡 Research Summary

The paper addresses the growing concern that the information and communication technology (ICT) sector accounts for a substantial share of global electricity consumption and carbon emissions. While prior work has explored renewable‑energy‑aware data centers, those approaches are limited by the concentration of large facilities in a few geographic locations, which reduces the likelihood of harvesting abundant green power that is highly dependent on local weather conditions. To overcome this limitation, the authors propose a novel, dual‑layer architecture that brings renewable energy generation and content storage down to the router level, thereby turning the network itself into a distributed, energy‑aware caching platform.

In the first layer, a distributed gradient‑based routing protocol continuously evaluates the amount of renewable energy (wind and solar) available at each router. Each router measures its own generation, advertises a “green metric” to its neighbors, and the routing algorithm preferentially selects paths that traverse routers with the highest renewable‑energy ratios. This approach is fully decentralized, requiring only lightweight metric exchanges, and it can adapt in real time to fluctuating weather patterns.

The second layer implements an Information‑Centric Networking (ICN) style in‑network caching mechanism. Once a green path has been identified, the system proactively pulls requested contents from the origin data center and stores them on the routers along that path. The caching policy builds on a basic Least‑Recently‑Used (LRU) scheme but is augmented with a dynamic adjustment that favors keeping objects on routers that currently have abundant renewable power. When a router’s green supply drops, cached items are evicted more aggressively, allowing the router (or its line cards) to be powered down, thereby reducing brown‑energy consumption.

A detailed energy model is introduced: router power consumption is split into chassis and line‑card components, and the model accounts for the possibility of powering off individual line cards. Renewable generation at a router, r_eP_C(X,t), is computed from real meteorological data (wind speed and global horizontal irradiance) for both wind turbines and solar panels. The brown‑energy consumption is the residual after subtracting renewable generation, and the overall brown‑energy reduction for the network (σ_n) and for the data centers (σ_d) are expressed as utility functions. A tunable parameter α∈