Renewable Energy-Aware Joint Caching and Routing for Green Communication Networks

• Published in: IEEE systems journal 2018-03, Vol.12 (1), p.768-777
• Main Authors: Khreishah, Abdallah, Bany Salameh, Haythem, Khalil, Issa, Gharaibeh, Ammar
• Format: Article
• Language: English
• Subjects: Approximation algorithms; Caching; Carbon; Communication networks; Computer simulation; distributed algorithms; Energy; Energy conservation; Energy consumption; Energy management; Environmental impact; Green products; Internet; joint caching and routing renewable energy; Linear programming; Mixed integer; network coding; Optimization; Polynomials; Renewable energy; Renewable energy sources; Renewable resources; Rounding; Routing; Shortest-path problems; User satisfaction; Video on demand
• ISSN: 1932-8184; 1937-9234
• DOI: 10.1109/JSYST.2016.2530695

The Internet is one of the fastest increasing contributors to carbon emission. Content distribution as video on demand constitutes the majority of the Internet traffic. In order to reduce the Internet's carbon footprint, we propose greener mechanisms for content delivery that utilize the use of renewable energy and content caching concept. If renewable energy is not enough to satisfy a given user's demand, we use brown energy to satisfy the demand. Specifically, we consider the joint routing and caching problem with the objective of minimizing the brown energy usage while satisfying the users' requests. We formulate the problem as a mixed integer-linear program (MILP) and prove that it is NP-hard. Accordingly, we present two relaxation techniques to find an efficient solution in a polynomial time (within 10% of the optimal). The first technique is based on relaxation and rounding. The other one is a near-optimal solution based on sequential fixing, where the binary variables are determined iteratively by solving a sequence of linear programs. Then, we develop a gradient-based distributed algorithm that can adapt to the changes in traffic and renewable energy conditions. Finally, we show that by utilizing network coding, the problem can be formulated using linear programming, which has polynomial-time complexity. Simulation results are provided, which verify the effectiveness of our optimization framework and demonstrate the significant energy saving achieved (up to 90%) over the nonenergy-aware shortest path-routing method.