Grant: Discovery Grants Program - Individual

Network slicing is an enabling technology for the future Internet and next-generation communication networks (5G and beyond 5G mobile networks). It transforms a network from a "one size fits all" architecture, to a new paradigm where network slices/virtual networks are created, with appropriate isolation, resources, and optimized topology to serve a particular application or service. Life-cycle management of network slices is important to meet applications' Quality of Service (QoS) requirements on a dynamic and evolving network. Slice life-cycle management involves three phases: i) instantiation, configuration, and activation phase to orchestrate slices, with diverse QoS requirements, possibly spanning multiple network segments; ii) run-time phase to perform slice operational and maintenance tasks such as, performance monitoring, failure management, and resource adaptation; and iii) decommissioning phase that re-optimizes resources after decommissioning network slice(s). Slice life-cycle management faces unique challenges due to incorporation of virtualization and software defined networking, and traditional management approaches fall short in ensuring best slice operational performance, desired resource efficiency, and rapid incident management. This research program aims at overcoming some of the challenges pertaining to the development of a life-cycle management system for 5G and beyond 5G network slices. Towards this goal, my students and I will pursue three main objectives. First, we will develop optimization models and scalable algorithms for instantiating slices with heterogeneous QoS requirements to minimize capital and operational expenses. Second, we will explore optimization and data-driven techniques to devise intelligent monitoring schemes, proactive and reactive failure management approaches, and predictive slice adaptation mechanisms. Failure management approaches will include machine learning based anomaly detection and mitigation as well as automated failure detection, localization, and identification models. Slice adaptation will benefit from predicting a set of slice QoS requirement changes and accommodating them exploiting statistical multiplexing gains. Third, we will devise re-optimization and consolidation algorithms to increase resource efficiency of the network infrastructure. We will evaluate our algorithms and models through empirical and testbed experiments, and validate them using real network data. The outcome of the research program will contribute to the development of a slice life-cycle management system and help Canadian network operators optimize resource utilization, reduce capital and operational expenditures, minimize number of network outages, and increase revenue. My students will develop optimization frameworks, algorithmic tools, and machine learning models and evaluate them through rigorous experiments, and accumulating skills along the way that will serve them well in industry or academia.