Residential short-term energy consumption forecasting plays an essential role in modern decentralized power systems. The rise of innovative prediction methods able to handle the high volatility of users’ electrical load has posed the basis to accomplish this task. However these methods, which mostly rely on Artificial Neural Networks, require that a huge amount of users’ fine-grained sensitive consumption data are centrally collected to train a generalized forecasting model, with implications on privacy and scalability. This paper proposes an innovative architecture specifically designed to overcome this need. By exploiting Federated Learning and Edge Computing capabilities, many Long Short-Term Memory (LSTM) models are locally trained by different users based on their own historical energy consumption samples. Such models are then aggregated by a specific-purpose node to build a generalized model that is re-distributed for improved forecasting at the edge. For better forecasting, our proposed local training procedure takes as input relevant features related to calendar (i.e., hour, weekday and average consumption of previous days) and weather conditions (i.e., clustered apparent temperature), and the architecture can group users according to consumption similarities (using K-means) or socioeconomic affinities. We thoroughly evaluate the approach through simulations, showing that it can lead to similar forecasting performance than a state-of-the-art centralized solution in terms of Root Mean Square Error (RMSE), but with up to an order of magnitude lower training time and up to 50 times less exchanged data when samples are recorded at finer granularity than one hour. Nonetheless, it keeps sensitive data local and therefore guarantees users’ privacy.

Virtualization technologies are changing the way network operators deploy and manage Internet services. In particular in this study we focus on the new Network Function Virtualization (NFV) paradigm, which consists in instantiating Virtual Network Function (VNFs) in Commercial-Off-The-Shelf (COSTS) hardware. Adopting NFV network operators can dynamically instantiate Network Functions (NFs) based on current demands and network conditions, allowing to save capital and operational costs. Typically, VNFs are concatenated together in a sequential order to form Service Chains (SCs) that provide specific Internet Services to the users. In this paper we study different approaches to provide the resiliency of SCs against single-link and single-node failures. We propose three Integer Linear Programming (ILP) models to solve the VNF placement problem with the VNF service chaining while guaranteeing resiliency against single-node/link, single-link and single-node failures. Moreover we evaluate the impact of latency of SCs on the VNFs distribution. We show that providing resiliency against both single-link and single-node failures necessitates the activation of twice the amount of resources in terms of nodes, and that for latency critical services providing resiliency against single-node failures comes at the same cost with respect to resiliency against single-link and single-nodes failures.

The Network Functions Virtualization (NFV) paradigm is the most promising technique to help network providers in the reduction of capital and energy costs. The deployment of virtual network functions (VNFs) running on generic x86 hardware allows higher flexibility than the classical middleboxes approach. NFV also reduces the complexity in the deployment of network services through the concept of service chaining, which defines how multiple VNFs can be chained together to provide a specific service. As a drawback, hosting multiple VNFs in the same hardware can lead to scalability issues, especially in the processing-resource sharing. In this paper, we evaluate the impact of two different types of costs that must be taken into account when multiple chained VNFs share the same processing resources: the upscaling costs and the context switching costs. Upscaling costs are incurred by VNFs multi-core implementations, since they suffer a penalty due to the needs of load balancing among cores. Context switching costs arise when multiple VNFs share the same CPU and thus require the loading/saving of their context. We model through an ILP problem the evaluation of such costs and we show their impact in a VNFs consolidation scenario, when the x86 hardware deployed in the network is minimized.

LoRa is a long-range and low-power radio technology largely employed in Internet of Things (IoT) scenarios. It defines the lower physical layer while other protocols, such as LoRaWAN, define the upper layers of the network. A LoRaWAN network assumes a star topology where each of the nodes communicates with multiple gateways which, in turn, forward the collected data to a network server. The main LoRaWAN characteristic is the central role of the gateways; however, in some application scenarios, a much lighter protocol stack, relying only on node capabilities and without the presence of gateways, can be more suitable. In this paper, we present a preliminary study for realizing a LoRa-based mesh network, not relying on LoRaWAN, that implements a peer-to-peer communication between nodes, without the use of gateways, and extends node reachability through multi-hop communication. To validate our investigations, we present a hardware/software prototype based on low-power-consumption devices, and we preliminarily assess the proposed solution.

Volumetric distributed Denial-of-Service (DDoS) attacks have become one of the most significant threats to modern telecommunication networks. However, most existing defense systems require that detection software operates from a centralized monitoring collector, leading to increased traffic load and delayed response. The recent advent of Data Plane Programmability (DPP) enables an alternative solution: threshold-based volumetric DDoS detection can be performed directly in programmable switches to skim only potentially hazardous traffic, to be analyzed in depth at the controller. In this paper, we first introduce the BACON data structure based on sketches, to estimate per-destination flow cardinality, and theoretically analyze it. Then we employ it in a simple in-network DDoS victim identification strategy, INDDoS, to detect the destination IPs for which the number of incoming connections exceeds a pre-defined threshold. We describe its hardware implementation on a Tofino-based programmable switch using the domain-specific P4 language, proving that some limitations imposed by real hardware to safeguard processing speed can be overcome to implement relatively complex packet manipulations. Finally, we present some experimental performance measurements, showing that our programmable switch is able to keep processing packets at line-rate while performing volumetric DDoS detection, and also achieves a high F1 score on DDoS victim identification.