In backbone networks, it is fundamental to quickly protect traffic against any unexpected event, such as failures or congestions, which may impact Quality of Service (QoS). Standard solutions based on Segment Routing (SR), such as Topology-Independent Loop-Free Alternate (TI-LFA), are used in practice to handle failures, but no distributed solutions exist for distributed and tactical congestion mitigation. A promising approach leveraging SR has been recently proposed to quickly steer traffic away from congested links over alternative paths. As the pre-computation of alternative paths plays a paramount role to efficiently mitigating congestions, we investigate the associated path computation problem aiming at maximizing the amount of traffic that can be rerouted as well as the resilience against any 1-link failure. In particular, we focus on two variants of this problem. First, we maximize the residual flow after all possible failures. We show that the problem is NP-Hard, and we solve it via a Benders decomposition algorithm. Then, to provide a practical and scalable solution, we solve a relaxed variant problem, that maximizes, instead of flow, the number of surviving alternative paths after all possible failures. We provide a polynomial algorithm. Through numerical experiments, we compare the two variants and show that they allow to increase the amount of rerouted traffic and the resiliency of the network after any 1-link failure.

Tactical Traffic Engineering (TE) solutions are a must to adapt traffic steering when unexpected congestions occur. While already available, centralized solutions to locally optimizes congested tunnels and links suffer from a slow reaction time of several minutes. To address this issue, we propose a distributed Congestion Mitigation (CM) mechanism that leverages Segment Routing (SR) to offload traffic away from congested links using Unequal Cost Multi Paths (UCMP) over alternative paths. In this paper, we introduce an efficient algorithm for alternative paths’ computation, and two methods to compute UCMP weights, depending on whether remote link loads are available or not. We show that the proposed path computation method is faster than a modified K-shortest path algorithm. For traffic splitting, we show that the knowledge of remote link loads and per-destination traffic is a key to mitigate congestions in loaded scenarios, approaching the results obtained with an optimal solution. However, when not available, a local solution can already mitigate congestions in lightly loaded scenarios.

This paper presents a study of the optimal Admission Control in Damper-based Networks (ACDN) problem. The use of dampers in large-scale networks is becoming increasingly beneficial for a wide range of applications as it provides a reliable means of achieving deterministic delay guarantees without the need for synchronization between routers. In this context, optimal admission control solutions are required to fully utilize capacity. The problem being studied is a variant of the Unsplittable Multi-Commodity Flow (UMCF) problem, with additional constraints related to forwarding and shaping. This paper proposes two Integer Linear Programming (ILP) formulations to address the ACDN problem. The former is a compact formulation, which is solved using the CPLEX solver. The latter is an extended path formulation, for which a Branch-and-Price algorithm is developed, including a column generation procedure, an efficient branching scheme, and reinforced by a primal heuristic. Tests on realistic instances show that solving the path formulation using the Branch-and-Price algorithm is better than solving the compact formulation using CPLEX. Our algorithm divides by 14 the average running time given by CPLEX, and the path formulation gives a stronger linear relaxation with an average optimally gap of 0.3%. This work builds upon previous research [1] that developed a heuristic for finding near-optimal solutions, and instead aims to find exact optimal solutions for ACDN.

This paper proposes an Attention-based Multi-Agent Cooperation (AMAC) approach to reduce message exchange overhead in Multi-Agent Reinforcement Learning-based smart load balancing. AMAC shares only most relevant messages across agents to coordinate decision-making without degrading original performance. Experiments show that AMAC significantly lowers inter-agent communications overhead and learning complexity and outperforms multiple MARL benchmarks in Key Performance Indicators (KPIs) and Key Quality Indicators (KQIs).

This paper explores the use of multi-agent deep learning as well as learning to cooperate principles to meet strict service level agreements, in terms of throughput and end-to-end delay, for a set of classified network flows. We consider agents built on top of a weighted fair queuing algorithm that continuously set weights for three flow groups: gold, silver, and bronze. We rely on a novel graph-convolution based, multi-agent reinforcement learning approach known as DGN. As benchmarks, we propose centralized and distributed deep Q-network algorithms and evaluate their performances in different network, traffic, and routing scenarios, highlighting both the effectiveness of our proposals and the importance of agent cooperation. We show that our DGN-based approach meets stringent throughput and delay requirements across different scenarios, decreasing silver and bronze flow median waiting delays by more than 50 % and reducing the SLA violations of the latter by nearly 60 %, with respect to a classic priority queuing approach.

In this paper, we explore the use of an actor-critic architecture for Deep Reinforcement Learning (DRL) to improve load balancing beyond traditional algorithms. Some centralized Reinforcement Learning (RL) algorithms have targeted in the reward function expression the Quality of Experience (QoE) for video flows, but this requires access to clients, or the Maximum Link Utilization (MLU) for other types of flows. In our approach, we tune the actor-critic algorithm to only leverage on QoS parameters in order to load balance traffic in the network and maximize the QoE experienced by the users. This avoids having to collect observations and performance measurements from client applications, as it only focuses on network metrics that can be easily measured. We explore both centralized and distributed solutions to assess the feasibility of the proposed smart load balancing solutions. We compare them to ECMP, QoE-based reward methods, and RILNET that uses an underlying DDPG optimization approach. The proposed algorithms are shown to outperform previous approaches.

With 5G networking, deterministic guarantees are emerging as a key enabler. In this context, we present a scalable Damper-based architecture for Large-scale Deterministic IP Networks (D-LDN) that meets required bounds on end-to-end delay and jitter. This work extends the original LDN [1] architecture, where flows are shaped at ingress gateways and scheduled for transmission at each link using an asynchronous and cyclic opening of gate-controlled queues. To further relax the need for clock synchronization between devices, we use dampers, that consist in jitter regulators, to control the burstiness flows to provide a constant target delay at each hop. We introduce in details how data plane functionalities are implemented at all nodes (gateways and core) and we derive how the end-to-end delay and jitter are calculated. For the control plane, we propose a column generation algorithm to quickly take admission control decisions and maximize the accepted throughput. For a set of flows, it determines acceptance and selects the best shaping and routing policy. Through a proof-of-concept implementation in simulation, we verify that the architecture meets promised guarantees and that the control plane can operate efficiently at large-scale.

Deterministic performance is a key enabler for 5G applications. While specific data-plane solutions have been proposed to reach a low deterministic end-to-end latency and jitter, legacy round-robin schedulers can already be used to guarantee bounds on the end-to-end latency, when associated with per-flow shapers. In this context, we propose a latency-guaranteed network slicing solution that trades-off between complexity and performance. We propose control plane algorithms to configure sub-channelized interfaces with an independent QoS scheduler at each physical port used by a slice. The algorithms allocate service rates and decide about queue assignments and routing inside each slice. Through numerical results on large network topologies, we demonstrate that our column-generation and two-steps algorithms can improve traffic acceptance while reducing the amount of reserved capacity.

Deterministic performance is a key enabler for 5G networking. In this context, we present a highly scalable Large-scale Deterministic Network (LDN) architecture providing end-to-end latency and bounded jitter guarantees in IP networks. At the data plane, flows are first shaped at ingress gateways using gate-control queues, achieving a very fine granularity compared to existing state of the art solutions. Inside the network, traffic is scheduled using an asynchronous cyclic queuing mechanism that can be implemented in real devices as it requires only 3 FIFO queues. The data plane relies on standard IP routing and a quasi-static mapping table to deterministically aggregate and forward packets over cycles with a low complexity in O(1). For the control plane, we present an advanced column generation algorithm to quickly take admission control decisions in large-scale networks. For a set of flows, it determines acceptance and selects the best shaping and routing policy. Through a proof-of-concept implementation and simulations, we show that our LDN architecture can guarantee end-to-end latency and bounded jitter. We also demonstrate that our advanced control plane algorithm brings an improvement up to 40% in terms of accepted traffic over classical routing.

The surge of video traffic is a challenge for service providers that need to maximize Quality of Experience (QoE) while optimizing the cost of their infrastructure. In this paper, we address the problem of routing multiple HTTP-based Adaptive Streaming (HAS) sessions to maximize QoE. We first design a QoS-QoE model incorporating different QoE metrics which is able to learn online network variations and predict their impact on representative classes of adaptation logic, video motion and client resolution. Different QoE metrics are then combined into a QoE score based on ITU-T Rec. P.1202.2. This rich score is used to formulate the routing problem. We show that, even with a piece-wise linear QoE function in the objective, the routing problem without controlled rate allocation is non-linear. We therefore express a routing-plus-rate allocation problem and make it scalable with a dual subgradient approach based on Lagrangian relaxation where subproblems select a single path for each request with a trivial search, thereby connecting explicitly QoE, QoE and HAS bitrate. We show with ns-3 simulations that our algorithm provides values for HAS QoE metrics (quality, rebufferings, variation) equivalent to MILP and better than QoS-based approaches.

We consider a multi‐domain network scenario and we study the Inter‐Domain Path Computation problem under the Domain Uniqueness constraint ( ‐ ), that is, a path cannot visit a domain twice. It is known that hierarchical Path Computation Element (h‐PCE) architecture, that is commonly used to solve ‐ , shows poor scalability with respect to the number of domains. For this reason, we devise a new domain clustering concept allowing one to artificially reduce the number of domains in an offline phase, in order to solve ‐ with lower complexity at run‐time. More specifically, we first prove the ‐completeness of the feasibility problem associated with ‐ and the inapproximability of ‐ itself. Yet, we show that the number of domains is the real computational bottleneck for the solution of ‐ . Then we provide a necessary and sufficient condition for a domain clustering to be proper, that is, without loss of optimality. Such a condition can be verified offline on the inter‐domain graph. We finally show via numerical experiments the impact of the inter‐domain treewidth on the computational speed‐up brought by proper clustering.

Content Delivery Networks (CDN) are witnessing the outburst of video streaming (e.g., personal live streaming or Video-on-Demand) where the video content, produced or accessed by mobile phones, must be quickly transferred from a point to another of the network. Whenever a user requests a video not directly available at the edge server, the CDN network must (1) identify the best location in the network where the content is stored, (2) set up a connection and (3) deliver the video as quickly as possible. For this reason, existing CDNs are adopting an overlay structure to reduce latency, leveraging the flexibility introduced by the Software Defined Networking (SDN) paradigm. In order to guarantee a satisfactory Quality of Experience (QoE) to users, the connection must respect several Quality of Service (QoS) constraints. In this paper, we focus on the sub-problem (2), by presenting an approach to efficiently compute and maintain paths in the overlay network. Our approach allows to speed up the transfer of video segments by finding minimum delay overlay paths under constraints on hop count, jitter, packet loss and relay node capacity. The proposed algorithm provides a near-optimal solution, while drastically reducing the execution time. We show on traces collected in a real CDN that our solution allows to maximize the number of fast video transfers.

Load-balancing and network optimization in SDN networks require efficient flow splitting during the path computation phase. The way flow splitting is typically implemented in switches is to map the output of an hash function computed on the headers of incoming flows to the content stored in a Ternary Content Addressable Memory (TCAM), a very efficient but scarce resource. Although a large TCAM budget means that the flow distribution can more accurately model a fractional ideal, the distribution of flow volume amongst the paths is constrained in reality to use only a limited number of TCAM rows. In this paper, we present a flow splitting algorithm that maximizes the total number of demands allocated in the network according to the TCAM size constraints and, at the same time, minimize the total routing cost. Although the problem is NP-hard, we show through simulations that we can achieve good approximations of the optimal solution in a reasonable amount of time.

Energy is classically considered as a critical resource in Wireless Sensor Networks (WSNs). These networks are composed of tiny devices that auto-organize around one or few gateways, which may have various roles from simple reference or traffic sinks to full network orchestrator. Such a gateway could influence the network behavior, for instance by decreasing activity when energy becomes scarce. It however needs to be able to estimate the nodes remaining energy. Indeed, this gateway is on the path of all traffic going in or out the WSN. This traffic sample could be used to acquire a coarse estimate of individual nodes energy consumption. The accuracy of this estimation can then be improved by explicit signaling if needed. This paper presents Tee, a set of such Traffic-based energy estimators that operates at the WSN gateway. We evaluate, by simulation, the accuracy of two such estimators in IEEE 802.15.4 networks running RPL and ContikiMAC, a duty cycled MAC layer. Results show that such silent estimators benefit from information already available at the gateway, such as the routing topology. However, they still underestimate the consumption due to the routing control messages, to the packets strobing, or to contention and collisions and can easily be complemented by lightweight explicit calibrations.

In Wireless Sensor Networks (WSNs) for monitoring applications, energy saving and fast data collection are two challenging tasks. Asynchronous radio duty cycling protocols can achieve very low energy consumption in low traffic conditions and they are fault-tolerant to clock drifts. However, they may exhibit a delay degradation due to the decoupled wake-up periods of the nodes. In this paper, we present RAWMAC, a cross-layer approach where RPL, a tree-based routing protocol, orchestrates the asynchronous duty-cycled ContikiMAC MAC layer. The wake-up instants of the nodes are dynamically aligned, with respect to the RPL topology, to minimize the delay for data collection. We implement RAWMAC for the Contiki operating system and we analyze the impact of several key system parameters. Results show that RAWMAC outperforms ContikiMAC in terms of delay for data collection, while keeping the same performance in terms of throughput and energy consumption.

The Internet of Things (IoT) aims at connecting billions of devices in an Internet-like structure. This gigantic information exchange enables new opportunities and new forms of interactions among things and people. A crucial enabler of robust applications and easy smart objects' deployment is the availability of mechanisms that minimize (ideally, cancel) the need for external human intervention for configuration and maintenance of deployed objects. These mechanisms must also be scalable, since the number of deployed objects is expected to constantly grow in the next years. In this work, we propose a scalable and self-configuring peer-to-peer (P2P)-based architecture for large-scale IoT networks, aiming at providing automated service and resource discovery mechanisms, which require no human intervention for their configuration. In particular, we focus on both local and global service discovery (SD), showing how the proposed architecture allows the local and global mechanisms to successfully interact, while keeping their mutual independence (from an operational viewpoint). The effectiveness of the proposed architecture is confirmed by experimental results obtained through a real-world deployment.

In scenarios like the surveillance of isolated areas, when the border node of a network does not have a permanent connection with the Internet, Wireless Sensor Networks (WSNs) are calling for resilient in-network data storage techniques which minimize the risk of data loss. The efficiency of these techniques can be largely improved exploiting information on the status of the network, such as that used by routing protocols. In particular, one of the most used protocol in Internet of Things (IoT) scenarios is the IPv6 Routing Protocol for Low power and lossy networks (RPL). In this paper, we propose a redundant distributed data storage and retrieval mechanism to increase the resilience and storage capacity of a RPL-based WSN against local memory shortage. We evaluate our approach in the Contiki operating system through extensive analysis with the Cooja simulator.

This paper addresses the problem of engineering energy-efficient target detection applications using unattended Wireless Sensor Networks (WSNs) for long-lasting surveillance of areas of interest. As battery energy depletion is an issue in this context, an approach consists of switching on and off sensing and communication modules of wireless sensors according to duty cycles. Making these modules work in an intermittent fashion impacts (i) the latency of notification transmission (depending on the communication duty cycle) and (ii) the probability of missed target detection (depending on the number of deployed nodes and the sensing duty cycle). In order to optimize the system parameters according to performance objectives, we first derive an analytical engineering toolkit which evaluates the probability of missed detection (Pmd), the notification transmission latency (D), and the network lifetime (¿) under the assumption of random node deployment. Then, we show how this toolbox can be used to optimally configure system parameters under realistic performance constraints.