As automation and digitalisation become more pervasive, understanding their social implications, opportunities and risks, as well as appropriate policies and regulatory needs, require increasing inter-disciplinary efforts. This paper describes the motivation, format and outcomes of the AIvolution event about AI and digital technologies, organized in the European Parliament (EP) on November 2023, for that purpose. It presents the main topics addressed, the key contributions that were presented, the discussions that took place, and the main takeaways, as well as the perspectives to continue these exchanges in further initiatives.

This paper considers decentralized systems of multiple agents that interact in order to perform distributed computing or efficient information transfer in variant, constrained environments. These systems appear in various relevant use cases, including reliable multicast distribution, or dynamic resource allocation in systems in the Internet edge (datacenters, IOT deployments). This paper introduces an abstract, mathematical model, that allows to study analytically the behavior of these systems, as a set of interacting requesters and providers. The paper describes system orbits, and concentrates on the study of single-orbit systems. Ergodicity of system behavior in the single-orbit case is proved, and a full description of the stationary system behavior is derived. Closed expressions of the stationary distribution of requester decisions are provided. Analytical results are validated through extensive simulations. These single-orbit results are a necessary step for the analysis and further optimization of dynamic performance of these systems.

Expert systems for fault diagnosis are computationally expensive to build and maintain, and lack scalability and inherent adaptability to unknown events or modifications in the topology of the monitored system. While data-driven feature selection mechanisms can facilitate diagnosis without the hardship of developing and maintaining expert systems, purely data-driven mechanisms lack understanding of semantic importance within a feature set, and would benefit from additional domain knowledge. Part of this additional knowledge can be extracted from metadata. The proposed approach combines data-driven metrics and semantic information contained in the feature names to produce selections of features which best represent an underlying event. This study extends a cross entropy based optimization method to join semantic importance with data behavior. A benchmarking architecture is introduced to evaluate the benefits of semantic analysis, and demonstrate the performance and robustness of semantic feature selection on different types of faults in network telemetry datasets, modeled with the YANG data modeling language. The results illustrate the interest of such a complementary meta-data analysis for data-driven fault diagnosis, and highlight the robustness of the studied approach against variations in the input feature set.

Cloud environments require dynamic and adaptive networking policies. It is preferred to use heuristics over advanced learning algorithms in Virtual Network Functions (VNFs) in production because of high-performance constraints. This paper proposes Aquarius to passively yet efficiently gather observations and enable the use of machine learning to collect, infer, and supply accurate networking state information - without incurring additional signaling and management overhead. This paper illustrates the use of Aquarius with a traffic classifier, an auto-scaling system, and a load balancer - and demonstrates the use of three different machine learning paradigms - unsupervised, supervised, and reinforcement learning, within Aquarius, for inferring network state. Testbed evaluations show that Aquarius increases network state visibility and brings notable performance gains with low overhead.

In order to dynamically manage and update networking policies in cloud data centers, Virtual Network Functions (VNFs) use, and therefore actively collect, networking state information - and in the process, incur additional control signaling and management overhead, especially in larger data centers. In the meantime, VNFs in production prefer distributed and straightforward heuristics over advanced learning algorithms to avoid intractable additional processing latency under high-performance and low-latency networking constraints. This paper identifies the challenges of deploying learning algorithms in the context of cloud data centers, and proposes Aquarius to bridge the application of machine learning (ML) techniques on distributed systems and service management. Aquarius passively yet efficiently gathers reliable observations, and enables the use of ML techniques to collect, infer, and supply accurate networking state information—without incurring additional signaling and management overhead. It offers fine-grained and programmable visibility to distributed VNFs, and enables both open- and close-loop control over networking systems. This paper illustrates the use of Aquarius with a traffic classifier, an auto-scaling system, and a load balancer—and demonstrates the use of three different ML paradigms—unsupervised, supervised, and reinforcement learning, within Aquarius, for network state inference and service management. Testbed evaluations show that Aquarius suitably improves network state visibility and brings notable performance gains for various scenarios with low overhead.

The purpose of network load balancers is to optimize quality of service to the users of a set of servers– basically, to improve response times and to reducing computing resources– by properly distributing workloads. This paper proposes a distributed, application-agnostic, Hybrid Load Balancer (HLB) that– without explicit monitoring or signaling– infers server occupancies and processing speeds, which allows making optimised workload placement decisions. This approach is evaluated both through simulations and extensive experiments, including synthetic workloads and Wikipedia replays on a real-world testbed. Results show significant performance gains, in terms of both response time and system utilisation, when compared to existing load-balancing algorithms.

This paper addresses the problem of high-quality packet pacing for constant-rate packet consumption systems, with strict buffering limitations. A mostly-software pacing architecture is developed, which has minimal hardware requirements, satisfied by commodity servers - rendering the proposed solution easily deployable in existing (data-centre) infrastructures. Two algorithms (free-running and frequency-controlled pacing, for explicitly and implicitly indicated target rates, respectively) are specified, and formally analysed. The proposed solution, including both algorithms, is implemented, and is tested on real hardware and under real conditions. The performance of these implementations is experimentally evaluated and compared to existing mechanisms, available in general-purpose hardware. Results of both exhaustive experiments, and of an analytical modeling, indicate that the proposed approach is able to perform low-jitter packet pacing on commodity hardware, being thus suitable for constant rate transmission and consumption in media production scenarios.

Edge computing technologies and integrated architectures have been deployed to accommodate Internet traffic growth. These architectures include facilities (cloudlets, micro DCs) to cache and serve contents close to consumers. Resulting systems adapt to observed request/consumption patterns by allowing cloudlet coordination for content caching/dissemination. This paper presents a novel analytical model of transient dynamics of the cloudlets set. The model is used to study system convergence, stability and delivered content locality. Results from this model are validated via simulations.

BIER (Bit-Indexed Explicit Replication) alleviates the operational complexities of multicast protocols (associated to the multicast tree and the incurred state in intermediate routers), by allowing for source-driven, per-packet destination selection, efficient encoding thereof in packet headers, and stateless forwarding along shortest-path multicast trees. BIER per-packet destination selection enables efficient reliable multicast delivery: packets not received by a subset of intended destinations can be efficiently BIER-retransmitted to only that subset. While BIER-based reliable multicast exhibits attractive performance attributes, relying on source retransmissions for packet recovery may be costly - even unnecessary, if topologically close peers are able to provide a copy of the packet. Thus, this paper extends the use of reliable BIER multicast to allow recovery also from peers, using Segment Routing (SR) to steer retransmission requests through a set of potential (local) candidates, before requesting retransmissions from the source as a last resort only. A general framework is introduced, which can accommodate different policies for the selection of candidate peers for retransmissions. Simple (both static and adaptive) policies are introduced and analyzed, both (i) theoretically and (ii) by way of simulations in data-center-like and real-world topologies. Results indicate that local peer recovery is able to substantially reduce the overall retransmission traffic, and that this can be achieved through simple policies, where no signalling is required to build a set of candidate peers.

In large low-power networks of battery-driven sensors, power outages are a major concern and communication rates have to be carefully designed in order to optimize energy consumption, network connectivity and sensors lifetime. In some IoT use cases, power can be supplied to sensors by way of renewable energy automatic harvesting (solar panels, etc.). Given the high variability of energy arrival processes, energy consumption in sensors, in particular caused by transmissions to the sink, has to be aligned with energy harvesting patterns, so as to maximize throughput while avoiding power outages that may arise when the battery is empty. This paper proposes ODMAC++, an extension to a well-known protocol for sensor transmission scheduling in a WSN. ODMAC++ relies on learning techniques to adapt sensors communication rate to energy harvesting patterns, and uses a beaconing mechanism whose frequency is adjusted based on past measurements on the harvested energy process. Simulations based on analytical energy arrival models and on real solar radiation measurements indicate that ODMAC++ is able to avoid power outages and to cope with battery limitation and energy variations due to variability in time.

The controller placement problem remains a key aspect of Software Defined Networking (SDN). The selection of a suboptimal controller may impact severely the performance of the control layer and, in consequence, cause a substantial degradation in the data layer. Different approaches for controller placement have been proposed with various objectives in mind, but most of them do not consider the characteristics of the resulting control layer in terms of robustness. In this paper we propose and formalize a complete metric for estimating robustness of a SDN control layer, and we evaluate two heuristics for controller selection and control layer construction: Fast Failover and a simplified version of k-Critical. The results of the performed evaluation indicate that the control layer topology induced by k-Critical is less prone to failures, more robust and more homogenous than those computed by Fast Failover.

Traffic between an edge network and the rest of the Internet can be represented as a dynamic loop-free graph. Understanding in depth the dynamics in time and space (spatial structure, topological breadth, destination persistency, traffic dominating paths) of this graph provides significant insight on the Internet internal architecture and capabilities. This paper analyzes inter-domain traffic from a large campus network based on one month by way of Netflow measurements. Our analysis reveals the topological properties and structure of the traffic graph (breadth, depth, volume), the stability of contacted destinations and the relationship between their popularity and their path length. Based on the observed traffic, we explore the suitability of a simple mathematical model to describe the structure of the outgoing traffic graph.

Jitter is used in wireless ad hoc networks to reduce the number of packet collisions and the number of transmissions. This is done by scheduling random back-off for each packet to be transmitted and by piggybacking multiple packets in a single transmission. This technique has been standardized by the IETF in RFC 5148. This paper investigates on the impact of the standardized jitter mechanism on network-wide packet dissemination - i.e. flooding, an important component for many protocols used today. A novel analytical model is introduced, capturing standard jitter traits. From this model is derived accurate characterization of the effects of jittering on flooding performance, including the additional delay for flooded packets on each traversed network interface, the reduction of the number of transmissions over each network interface, and the increased length of transmissions, depending on jitter parameters. This paper also presents an analysis of the use of jitter in practice, over an 802.11 wireless link layer based on CSMA. The analytical results are then validated via statistical discrete event simulations. The paper thus provides a comprehensive overview of the impact of jittering in wireless ad hoc networks.

As wireless mesh networks are deployed, a new concept emerges: compound internetworks, i.e., internetworks that contain both wired networks and wireless mesh networks. Routing is one of the key challenges that arises in compound internetworks: indeed, while specific routing protocols are typically used for wired networks on one hand, and for wireless mesh networks on the other hand, it has been observed that operating a single routing protocol to manage a compound internetwork as a whole brings several advantages. In this realm, the IETF has thus standardized protocol extensions to Open Shortest Path First (OSPF, the routing protocol used by more than 50 % of the wired routers in today's Internet), enabling OSPF to operate simultaneously on wired networks, and on wireless mesh or moderately mobile ad hoc networks (MANETs). This paper evaluates the performance of OSPF coupled with such a protocol extension for MANETs on an experimental compound internetwork testbed. This paper reports on experiments carried out with OSPF operating simultaneously over Ethernet and 802.11b. Despite the limitations of the testbed, these experiments provide both a proof-of-concept and complementary results compared to prior work in the domain, which was mostly based on simulations, and focused on wireless ad hoc network scenarios only.

In practice, ad hoc networks are still too unreliable for standard mobile and vehicular communications. It is thus important to complement current protocols in this context, with schemes guaranteeing the exchange of critical data when needed. A promising approach in this realm is to use an overlay subgraph, over which critical messages are exchanged and acknowledged in a peer to peer fashion. Overlay nodes' local databases remain thus synchronized over time, at least concerning critical data. This paper elaborates on the problem of performance, related to the discovery and maintenance of such overlay networks in a mobile ad hoc context. We analyze SLOT, an overlay selected based on a Relative Neighbour Graph (RNG) scheme. We then apply SLOT to a standard IP protocol: OSPF, a popular routing protocol which has recently been extended, with RFC 5449 and RFC 5614, to work also on mobile ad hoc networks, and which makes use of a similar overlay synchronization subgraph. This paper compares the performance of these existing OSPF mechanisms with that of SLOT-OSPF, a novel OSPF extension for mobile ad hoc networks using SLOT. Simulations show that SLOT-OSPF produces drastically less control traffic than RFC 5449 or RFC 5614, allowing SLOT-OSPF to function correctly while the other existing approaches stall, when the number of routers in the domain is large.