Robust Proactive Mobility Management in Named Data Networking under Erroneous Content Prediction

Abstract

Named Data Networking (NDN) is a promising paradigm for the future Internet to survive the growing data demand. Supporting seamless operation during user mobility is one of the main challenges in NDN. In this paper, we investigate optimal caching for producer mobility under prediction uncertainties. Mainly, we propose a stochastic optimization framework that exploits location and data requests' predictors to cache data proactively before handover. We model the problem using Chance Constraint Programming (CCP) that probabilistically incorporates the uncertainty in data prediction and models the trade-off between network overhead and Consumer satisfaction. A deterministic formulation is derived to obtain a closed form Integer Linear Programming model based on the prediction error model. The proposed framework is then implemented in ndnSIM and Gurobi, and simulation experiments are conducted to provide benchmark solutions for robust proactive caching. The results show that such robust scheme satisfies the consumers' quality of experience under imperfect prediction of future content requested from mobile producers. Hence, sustains the prediction gains over conventional non- predictive schemes without compromising the network overhead. We believe that such results drive incentives for deploying proactive mobility management in future NDN.