This paper addresses the system design for virtual network services, particularly focusing on virtual machine (VM) services. The VM service requires users to define a substantial number of specifications, such as CPU resources and latency bounds, as a prerequisite. This process is challenging for nonprofessional users and inefficient even for professional users. Users need to clearly understand the specifications they need and report their requirements to the service manager, which demands a high level of knowledge and experience of a system engineer. To make the system more accessible to users, we propose a framework that enables interaction with the virtual network service through natural language (NL) inputs from the users. The framework employs NL models to interpret user requests in NL format into service specifications to determine the specificationdependent optimal virtual network allocation. We demonstrated the effectiveness of the proposed framework through numerical experiments, which show that the user requests in NL are accurately interpreted and incorporated into the virtual network allocation.

To achieve carbon neutrality in telecommunication networks, this paper focuses on a virtual network (VN) allocation problem for telecommunication networks powered by renewable energy (RE). One of the most critical issues is the fluctuation of RE power caused by uncertain weather conditions. Under the fluctuation of RE power, we must allocate multiple VNs so that RE power is efficiently used while avoiding waste. To deal with the fluctuation problem of RE power, we model RE power as a stochastic variable that follows a certain probability distribution. However, there are still challenges in accurately estimating the true distribution due to the limited size and nonstationarity of RE power data. To this end, we propose a distributionally robust VN allocation model that minimizes the expectational cost of the total excess power for the worst-case probability distribution in the set of all possible distributions that generate actual RE power, i.e., the uncertainty set. The uncertainty set is defined on the basis of the Wasserstein ball, in which the center is an empirical distribution constructed from the accumulated RE data and the radius is a certain Wasserstein metric. Then, the true probability distribution is assumed to be in the uncertainty set. Finally, through numerical experiments, the effectiveness of the proposed method is shown even if the estimation of RE power distribution is uncertain due to the small size of RE power data.

To achieve carbon neutrality in telecommunications networks, the introduction of renewable energy is expected to further advance in the future. However, renewable energy such as solar power generation can experience large fluctuations in power output depending on weather conditions. Therefore, depending on the location and time period, there may be places where renewable energy cannot be fully consumed and surplus power is generated, and there may be places where renewable energy is not sufficient to meet the power demand. Toward network carbon neutrality, new methods need to be established from a network perspective that avoid inefficient power management problems caused by fluctuations in renewable energy. To solve this problem, we have been studying the control of power consumption locations and time periods by allocating workloads using virtualization technology. In this paper, we propose a workload scheduling method to increase renewable energy use. The problem to maximize the renewable energy use is formulated, and to solve it in a short time, a heuristic search method is proposed. The simulation results show that the proposed control method can increase the amount of renewable energy use while having tradeoffs with communication quality and equipment efficiency.

In this paper, we consider a physical network powered by renewable energy resources and a virtual network (VN) of a client service, which is composed of a client node, a virtual machine (VM) node, and a virtual link. Then, a robust VN allocation problem is formulated for multiple client services: given the location of client nodes, find the allocation of VM nodes and virtual links to maintain robustness for the uncertainty of traffic demands and renewable energy power; i.e., their prediction error. Specifically, we propose two robust allocation models: robust VN allocation model and two-stage robust VN allocation model, which are formulated on the basis of robust optimization and two-stage robust optimization, respectively. To show the effectiveness of two robust proposed models, we conduct numerical experiments under various prediction error patterns of traffic demands and renewable energy power. When prediction errors are large, the two proposed models acquire better average and worst-case performance than a deterministic model that does not handle prediction errors. In addition, we observe some patterns where the two-stage robust VN allocation model acquires better average performance than the robust VN allocation model instead of deteriorating the worst-case performance.