Open RAN (Radio Access Network) is revolutionizing the telecom space by introducing a framework based on the concepts of virtualization and openness. O-RAN fosters virtualized and disaggregated RAN components connected via open interfaces based on specifications by the O-RAN Alliance. The network is optimized using RAN intelligent controllers (RICs), which can take data-driven, closed-loop actions in a RAN built in a multi-vendor, interoperable environment. The goal of this paper is to provide insights and guidance about the paradigm shift brought by O-RAN in order to create open, softwarized, intelligent and optimized networks. We focus on the intelligence aspects by providing an in-depth view of the near-RT and non-RT RICs specified by the O-RAN Alliance, including the architecture and interfaces. A novel aspect of this paper is that we provide guidelines in terms of the artificial intelligence and machine learning (AI/ML) approaches and frameworks that are useful in the O-RAN context, and consider the applications (xApps and rApps) that can be created to programmatically and autonomously control and optimize the network through the RICs for V2X, Industry 5.0, and other very demanding service types. Additionally, we provide the E2E network slice orchestration architecture, and demonstrate the suitability of O-RAN for the requirements of the service types to be achieved. Finally, we discuss research challenges and opportunities and overview existing experimental research platforms that are used to innovate and drive advances in the O-RAN effort.
23 March 2020 in IEEE Transactions on Green Communications and Networking
In this study, a public parking-lot is assumed to schedule the charging of Electric Vehicles (EVs). Each EV owner upon arriving gives the energy demand as well as departure time to the system and immediately receives feedback; fulfilling or adjusting the request. The system designed in this study decides based on the previously admitted requests and the uncertain future demands in both Admission Control (AC) and Charge Scheduling (CS) mechanisms. We formulate a multi-stage stochastic programming model to minimize the expected total energy costs over the finite time horizon. Next, we approximate the model using a finite scenario tree. However, this model is computationally intractable, even for a moderate number of stages. Therefore, we customize a well-known decomposition procedure, Stochastic Dual Dynamic Programming (SDDP), to be matched the time-dependent charging conditions. Since the procedure takes several hours to obtain a high-quality solution, we run it once in the offline mode and employing the results for the online mode. The simulation results indicate that the proposed method outperforms the myopic approach, and obtains a close solution to the theoretical optimal value in terms of total costs and rejected demands.
5 February 2020 in IEEE Journal on Selected Areas in Communications
We study a general online combinatorial auction problem in algorithmic mechanism design. A provider allocates multiple types of capacity-limited resources to customers that arrive in a sequential and arbitrary manner. Each customer has a private valuation function on bundles of resources that she can purchase (e.g., a combination of different resources such as CPU and RAM in cloud computing). The provider charges payment from customers who purchase a bundle of resources and incurs an increasing supply cost with respect to the totality of resources allocated. The goal is to maximize the social welfare, namely, the total valuation of customers for their purchased bundles, minus the total supply cost of the provider for all the resources that have been allocated. We adopt the competitive analysis framework and provide posted-price mechanisms with optimal competitive ratios. Our pricing mechanism is optimal in the sense that no other online algorithms can achieve a better competitive ratio. We validate the theoretic results via empirical studies of online resource allocation in cloud computing. Our numerical results demonstrate that the proposed pricing mechanism is competitive and robust against system uncertainties and outperforms existing benchmarks.
Network management faces the interrelated challenges of increasing network complexity, meeting sophisticated business requirements, and being subject to human oversight. Self-driving networks possess the key properties to overcome such challenges. We present and implement a management system that addresses several elements of a self-driving network. Our system leverages intents, a policy-based paradigm and autonomic control loops. Intent-based networking allows us to formalize how an intent can be provided as input to a control loop, and how the complexities can be abstracted from the user. To realize and assure the intent, autonomic networking enables us to create Monitor-Analyze-Plan-Execute (MAPE) loops. Finally, we execute the control loops using a policy-based approach. We propose a policy abstraction to support requirements at different levels of abstraction, and an Application Programming Interface (API) layer to reduce management complexity from the user perspective. We propose a formal policy information model to model policies across layers of abstractions and to support simplified mapping and strong consistencies among various policy abstraction levels. We have implemented our proposal and present a proof-of-concept use-case to showcase the intent refinement.
This paper presents an online anomaly detection system capable of handling operational network traffic of large networks (such as an ISP). We also aim for an effective and practical diagnosis of anomalies diagnosis to produce actionable intelligence that enables automated response. To achieve these objectives, we use the following approaches. (1) We model the status of the network by a stream of tensors where each tensor cell contains a time series. (2) We detect anomalous tensors at discrete time steps using an unsupervised tensor representation learning model. (3) We produce actionable intelligence by diagnosing anomaly detection results and identifying the abnormal time series that are the most likely causes of each anomaly in the tensor. (4) We further analyze the traffic corresponding to each anomalous time series by an innovative method that extracts and isolates the attack traffic. (5) We provide solutions for streaming data anomaly detection challenges such as large volume, high velocity, seasonality, and concept drift. We apply our approach to the complete test set of UGR data to show its practicality and effectiveness. Not only can we detect and isolate most of the labelled attack traffic, but we also identify many organic attack activities in the UGR data. Our results on the complete UGR dataset show high detection and isolation rates for the labelled attacks in the dataset. We also report on additional organic attacks we detected that were originally labelled as background in the dataset. Our analysis shows that the isolated background traffic represents interesting and potentially malicious behavior and can provide invaluable insight for cyber-threat researchers.