Container technology plays an important role in the virtualization landscape today. A container uses its own file system consisting of a stack of layers, which are stored on the execution server's disk. Containers running on the same server share the layers they have in common, and this sharing results in valuable savings in server storage space. Many containers can run on a single server, but when their resource demands grow enough, they are distributed across a cluster of nodes/servers by orchestration systems, such as Kubernetes. In this work, we found that for the same amount of containers to run, the storage required is higher for clusters consisting of a larger number of nodes. The severity of this storage overhead depends on the scheduling policy used to select the nodes that run the containers. By comparing different storage-saving scheduling policies that differ from each other in the depth of storage knowledge they leverage to make decisions, our analysis reveals that only deep, layer-level knowledge can effectively counter the growth in storage demand as the cluster size increases. Policies with coarser-grained knowledge achieve limited benefit because they achieve performance that is nearly equal to that of a random, zero-knowledge policy.

Many cloud providers offer IoT services that simplify the collection and processing of IoT information. However, the IoT infrastructure composed of sensors and actuators that produces this information remains outside the cloud; therefore, application developers must install, connect and manage the cloud. This requirement can be a market barrier, especially for small/medium software companies that cannot afford the infrastructural costs associated with it and would only prefer to focus on IoT application developments. Motivated by the wish to eliminate this barrier, this paper proposes a Cloud of Things platform, called VirIoT, which fully brings the Infrastructure as a service model typical of cloud computing to the world of Internet of Things. VirIoT provides users with virtual IoT infrastructures (Virtual Silos) composed of virtual things, with which users can interact through dedicated and standardized broker servers in which the technology can be chosen among those offered by the platform, such as oneM2M, NGSI and NGSI-LD. VirIoT allows developers to focus their efforts exclusively on IoT applications without worrying about infrastructure management and allows cloud providers to expand their IoT services portfolio. VirIoT uses external things and cloud/edge computing resources to deliver the IoT virtualization services. Its open-source architecture is microservice-based and runs on top of a distributed Kubernetes platform with nodes in central and edge data centers. The architecture is scalable, efficient and able to support the continuous integration of heterogeneous things and IoT standards, taking care of interoperability issues. Using a VirIoT deployment spanning data centers in Europe and Japan, we conducted a performance evaluation with a two-fold objective: showing the efficiency and scalability of the architecture; and leveraging VirIoT’s ability to integrate different IoT standards in order to make a fair comparison of some open-source IoT Broker implementations, namely Mobius for oneM2M, Orion for NGSIv2, Orion-LD and Scorpio for NGSI-LD.

The integration of localization capabilities within the cellular architecture through dedicated 5G network functions has notably enhanced cellular positioning accuracy and enabled new location-based services. However, this architectural shift requires placing measurement acquisition and computation at the network edge and core, resulting in distributed computational resources and increased latency and security risks. As a result, minimizing latency, ensuring resilience against security threats, and achieving high accuracy become critical performance indicators for location-based services. This paper examines both 3GPP-standardized and O-RAN-based 5G architectures, detailing the key functions, interfaces, and parameters influencing the localization process, from measurement acquisition to position estimation. We define performance indicators to evaluate localization services and develop a system model that quantifies the costs associated with latency, accuracy, computation, and resilience to security threats. By jointly considering these factors, we formulate a multi-objective optimization problem that guides the selection of an optimal system configuration to simultaneously satisfy multiple localization requirements. We validate our approach through a case study of an end-to-end 5G system using both simulations and experimental data. Specifically, we evaluate various algorithms and implementations across standardized channels and scenarios. Furthermore, we conduct experimental measurements using Software-Defined Radios (SDRs) and open-source 5G platforms to assess operational latency with commercial-off-the-shelf (COTS) devices.

FPGA accelerators on the NIC enable the offloading of expensive packet processing tasks from the CPU. However, FPGAs have limited resources that may need to be shared among diverse applications, and programming them is difficult. We present a solution to run Linux's eXpress Data Path programs written in eBPF on FPGAs, using only a fraction of the available hardware resources while matching the performance of high-end CPUs. The iterative execution model of eBPF is not a good fit for FPGA accelerators. Nonetheless, we show that many of the instructions of an eBPF program can be compressed, parallelized or completely removed, when targeting a purpose-built FPGA executor, thereby significantly improving performance. We leverage that to design hXDP, which includes (i) an optimizing-compiler that parallelizes and translates eBPF bytecode to an extended eBPF Instruction-set Architecture defined by us; a (ii) soft-CPU to execute such instructions on FPGA; and (iii) an FPGA-based infrastructure to provide XDP's maps and helper functions as defined within the Linux kernel. We implement hXDP on an FPGA NIC and evaluate it running real-world unmodified eBPF programs. Our implementation is clocked at 156.25MHz, uses about 15% of the FPGA resources, and can run dynamically loaded programs. Despite these modest requirements, it achieves the packet processing throughput of a high-end CPU core and provides a 10x lower packet forwarding latency.