Grant: Discovery Grants Program - Individual

Communication service providers (CSPs) aim to transition from the human-driven to zero-touch service management to achieve complete autonomy in cloud-based telco-networks. The idea is to specify the service objectives (e.g., latency and throughput requirements) as intents and use closed-control loops and artificial intelligence (AI) to achieve them. However, to realise such envisioned autonomy in telco-networks, it is mandatory to continuously monitor the physical and virtual resources (virtual machines (VM) and containers) provisioned to deploy the components providing network services (virtual/cloud-native network functions (VNFs/CNFs)). The monitoring output is used to identify or predict the events which lead to physical resource contention causing performance degradation among co-located virtual resources fulfilling the intent. Information about such events is further used by the AI-based service controllers to timely initiate the remediation and optimization measures by performing internal service and resource reconfiguration. This is to fulfil the service objectives specified as intents without any human intervention. This research program aims to propose novel architectural frameworks and solutions for integrated self-management (self-monitoring, self-healing and self-optimizing) of telco-cloud resources to mitigate the service performance degradation because of physical resource contention while safeguarding the fulfilment of intents. This is to contribute towards the long-term vision of this research program which is to provide a vital breakthrough in research and scientific foundation making future generation telco-networks self-driving (autonomous) without human-control. The specific short-term objectives are to develop (i) a systematic benchmarking methodology to collect, measure and quantify the relevant system key performance indicators (KPIs) to model system's behaviour representing resource contention, (ii) single-loop feedback controllers at physical machine (PM) level to regulate resource contention by adjusting the resources inside each PM, and (iii) site-level and domain-level multi-loop feedback controllers to migrate VMs/containers between PMs to assist PM-level controllers to regulate resource contention. The expected outcomes are (i) a set of prediction procedures to estimate the performance degradation in VMs/containers, (ii) collection of PM-level, site-level and domain-level controllers regulating the resource contention in telco-cloud environment, and (iii) a comprehensive list of guidelines to help both performance-control systems and system engineers to make calculative decisions for reducing resource contention and increasing intent assurance. It is anticipated that the proposed research program will not only generate significant scientific, technological, and social impacts but will also contribute to the training of HQPs in research and development for future generation communication technology.