Smart manufacturing relies on the digitization of all the industrial processes, from production to business operations.It uses Industrial Internet of Things (IIoT) principles to equip devices with smart sensors and actuators, integrating machines and software through data collection, advanced computational methods, and remote control.Our research is motivated by a real digital transition application in the luxury fashion in Italy.The customers wish to update legacy systems, to comply with new Industry 4.0 standards.Due to industrial property requirements, as well as brand secrets, they require the whole architecture to run on-premises.A further requirement is that the system installation must be non-invasive, potentially running on systems with frugal setups in terms of hardware and software.Adhering to such requirements and principles, this paper proposes an architecture for data pipeline in smart manufacturing that runs on-premises, offering support to legacy machines.It is capable of identifying unknown hardware, in terms of semantics of its sensors.The core component of such a concrete architecture is an innovative Extract-Transform-Load (ETL) connector, called sEmantic eXtended ETL (exETL), that manages numerous heterogeneous data sources, and recognizes and configures automatically new machinery sensors.It employs a dedicated Machine Learning (ML) pipeline.The flexibility of the proposed architecture is compared to alternative solutions that exploit existing technologies.Its computational effectiveness is assessed by building an emulated environment, and running extensive experiments on real data.Our results show that our data pipeline is lightweight, more flexible than competitors, and capable of integrating legacy or new machinery seamlessly.

Latency-sensitive applications, such as autonomous driving in smart cities and smart industries, require a networking and computing infrastructure to support their operations. Cloud-to-edge continuum represents a promising architecture to provide computational capability close to edge devices. However, deploying latency-sensitive applications in the continuum is challenging due to the heterogeneity and the geographical distribution of the computing nodes. In this paper, we address the deployment problem in a tele-operated autonomous driving scenario, formulating the orchestration task as a Virtual Network Function Placement Problem (VNFPP) with multi-tier performance levels, enabling vertical scaling of computational resources per microservice. Our MILP model, MORAL, minimizes node centrality-based deployment costs while satisfying resource and end-to-end latency constraints. We tested our approach through extensive simulations on realistic network topologies and synthetic applications, showing that the proposed model improves deployment feasibility, latency compliance, and resource efficiency compared to single performance tier versions and baseline strategies.

The fifth generation (5G) of mobile network offers a remarkable degree of flexibility to mobile operators, enabling them to provide users with effective and tailored network services. Software Defined Networking (SDN), Network Function Virtualization (NFV), and edge computing have given the operator the opportunity to easily bring computational capacity to the edge and to support latency-sensitive services. While 5G standards have defined the technological and architectural frameworks to orchestrate services, finding effective resources management and QoS optimization policies is still an open research issue. In this paper, we propose an online orchestration methodology for a multi-user edge service. The orchestrator goal is to simultaneously maximize the QoS, and minimize the amount of resources needed. We provide a mathematical formulation to compute an optimal offline policy and derive an online approach based on a model-free Deep Reinforcement Learning (DRL) framework. As a novel feature, the DRL agent action is modeled as a parametric combinatorial problem. A tailored multi-objective reward function leads the agent towards an effective choice of parameters for such a model. Our models are built, trained and fine-tuned by exploiting real data. Extensive simulations in diverse scenarios show that our DRL online approach produces solutions with small gaps to the optimal offline ones, enabling the operator to both save resources and grant the users an adequate QoS level.

We tackle an optimization problem arising in the design of sensor networks: given a set of sensors, only one being connected to a backbone, to establish connection routes from each of them to the sink. Under a shortest path routing protocol, the set of connections form a spanning tree. Energy is required to transmit and receive data, and sensors have limited battery capacity: as soon as one sensor runs out of battery, a portion of the network is disconnected. We, therefore, search for the spanning tree maximizing the time elapsed before such a disconnection occurs, and therefore, maintenance is required. We propose new mathematical formulations for the problem, proving and exploiting theoretical results on its combinatorial structure. On that basis, we design algorithms offering a priori guarantees of global optimality. We undertake an extensive experimental campaign, showing our algorithms to outperform previous ones from the literature by orders of magnitude. We also identify which instance features have higher impact on network lifetime.

Multi-access Edge Computing (MEC) technology has emerged to overcome traditional cloud computing limitations, challenged by the new 5G services with heavy and heterogeneous requirements on both latency and bandwidth. In this work, we tackle the problem of clustering access points in MEC environments, introducing a set of clustering models to be deployed at the pre-provisioning phase. We go through extensive simulations on real-world traffic demands to evaluate the performance of the proposed solutions. In addition, we show how MEC hosts capacity violation can be decreased when integrating access points clustering into the orchestration model, by investigating on solution accuracy when applied on held-out users traffic demands. The obtained results show that our approach outperforms two state-of-the-art algorithms, reducing both memory usage and execution time, by 46% and 50%, respectively, in comparison to a baseline algorithm. It surpasses the two methods in gaining control over MEC hosts capacity usage for different maximum achieved occupancy levels on MEC hosts.

We consider a combinatorial optimization problem for spatial information cloaking. The problem requires computing one or several disjoint arborescences on a graph from a predetermined root or subset of candidate roots, so that the number of vertices in the arborescences is minimized but a given threshold on the overall weight associated with the vertices in each arborescence is reached. For a single arborescence case, we solve the problem to optimality by designing a branch-and-cut exact algorithm. Then we adapt this algorithm for the purpose of pricing out columns in an exact branch-and-price algorithm for the multiarborescence version. We also propose a branch-and-price-based heuristic algorithm, where branching and pricing, respectively, act as diversification and intensification mechanisms. The heuristic consistently finds optimal or near optimal solutions within a computing time, which can be three to four orders of magnitude smaller than that required for exact optimization. From an application point of view, our computational results are useful to calibrate the values of relevant parameters, determining the obfuscation level that is achieved. The online supplement is available at https://doi.org/10.1287/ijoc.2018.0813 .

This paper deals with the multiple vehicle balancing problem (MVBP). Given a fleet of vehicles of limited capacity, a set of vertices with initial and target inventory levels and a distribution network, the MVBP requires to design a set of routes along with pickup and delivery operations such that inventory is redistributed among the vertices without exceeding capacities, and routing costs are minimized. The MVBP is NP‐hard, generalizing several problems in transportation, and arising in bike‐sharing systems. Using theoretical properties of the problem, we propose an integer linear programming formulation and introduce strengthening valid inequalities. Lower bounds are computed by column generation embedding an ad‐hoc pricing algorithm, while upper bounds are obtained by a memetic algorithm that separate routing from pickup and delivery operations. We combine these bounding routines in both exact and matheuristic algorithms, obtaining proven optimal solutions for MVBP instances with up to 25 stations.

In this paper we face a very fundamental problem in Operations Research: to find good dual bounds to generic mixed integer mathematical programs (MIPs) as quickly as possible. In particular, we focus on the scenario where large scale data needs to be considered, multicore CPU architectures are available, and massive parallelism can be exploited by means of decomposition methods. We consider column generation techniques to solve extended formulations obtained by means of Dantzig-Wolfe decomposition for MIPs. We propose a concurrent algorithm, that relaxes the synchronized behavior of classical column generation. Our approach relies on simple data structures and efficient synchronization, still providing the same global convergence properties of classical sequential column generation methods. We present and discuss the results of an extensive experimental campaign, comparing our concurrent algorithm to both a naive parallelization of column generation and the cutting planes algorithm implemented in state-of-the-art commercial optimization packages, considering large scale datasets of a hard packing problem from the literature as representative benchmark. Our approach turns out to be on average one order of magnitude faster than competitors, attaining almost linear speedups as the number of available CPU cores increases.

One of the primary challenges associated with network functions virtualization (NFV) is the automated management of the service lifecycle. In this paper, we present a full software-based management and orchestration (MANO) stack which operates with OpenStack and OpenDaylight controllers and has the in-built functionality to automate the key phases of the NFV service lifecycle, namely resource discovery and matching, service mapping, service deployment, and monitoring. The MANO stack is being implemented by the EU FP7 project T-NOVA, with the components being released as open-source software. Service mapping and service deployment solutions developed in the scope of T-NOVA are presented in detail. As a proof-of-concept, we evaluate the performance of a virtualized traffic classifier network function, demonstrating the gains of virtualized hardware acceleration.

Major interest is currently given to the integration of clusters of virtualization servers, also referred to as `cloudlets' or `edge clouds', into the access network to allow higher performance and reliability in the access to mobile edge computing services. We tackle the edge cloud network design problem for mobile access networks. The model is such that the virtual machines (VMs) are associated with mobile users and are allocated to cloudlets. Designing an edge cloud network implies first determining where to install cloudlet facilities among the available sites, then assigning sets of access points, such as base stations to cloudlets, while supporting VM orchestration and considering partial user mobility information, as well as the satisfaction of service-level agreements. We present link-path formulations supported by heuristics to compute solutions in reasonable time. We qualify the advantage in considering mobility for both users and VMs as up to 20% less users not satisfied in their SLA with a little increase of opened facilities. We compare two VM mobility modes, bulk and live migration, as a function of mobile cloud service requirements, determining that a high preference should be given to live migration, while bulk migrations seem to be a feasible alternative on delay-stringent tiny-disk services, such as augmented reality support, and only with further relaxation on network constraints.

Major interest is currently given to the integration of clusters of virtualization servers, also referred to as `cloudlets', into the access network to allow higher performance and reliability in the access to mobile cloud services. We tackle the cloudlet network design problem for mobile access networks. The model is such that virtual machines are associated with mobile users and are allocated to cloudlets. Designing a cloudlet network implies first determining where to install cloudlet facilities among the available sites, then assigning sets of access points such as base-stations to cloudlets, while supporting virtual machine migrations and taking into account partial user mobility information, as well as the satisfaction of service-level agreements. We present link-path formulations supported by heuristics to compute solutions in reasonable time. We qualify the advantage in considering mobility for both users and virtual machines as up to 40% less cloudlet facilities to install and 40% less virtual machine migrations to execute. We compare two migration modes, bulk and live migration, as a function of mobile cloud service requirements, determining that a high preference should be given to bulk migrations for delay-stringent services such as augmented reality support, while for applications with less stringent delay requirements, live migration appears as largely preferable.

In this article, we consider a variation of the vehicle routing problem arising in the optimization of waste management systems. Constraints imposing adequate level of service to the citizens and even workload among the drivers make the problem challenging and ask for the design of specialized algorithmic approaches. We propose an exact optimization algorithm, in which dynamic generation of rows and columns is done in a branch‐and‐bound framework; exact and heuristic algorithms are proposed for the pricing problem. Experimental tests on data‐sets from the literature show that our algorithm outperforms previous ones and it is able to solve instances of realistic size to proven optimality in reasonable computing time. © 2014 Wiley Periodicals, Inc. NETWORKS, Vol. 64(4), 282–291 2014

Given n points in ℝd and a maximum allowed tolerance ϵ > 0, the minimum hyperplanes clustering problem consists in finding a minimum number of hyperplanes such that the Euclidean distance between each point and the nearest hyperplane is at most ϵ. We present a column generation approach for this problem based on a mixed integer nonlinear formulation in which the master is a set covering problem and the pricing subproblem is a mixed integer program with a nonconvex normalization constraint. We propose different ways of generating the initial pool of columns and investigate their impact on the overall algorithm. Since the pricing subproblem is substantially complicated by the ℓ2-norm constraint, we consider approximate pricing subproblems involving different norms. Some strategies for refining the solution and speeding-up the overall method are also discussed. The performance of our column generation algorithm is assessed on realistic randomly generated instances as well as on real-world instances.

Abstract In many telecommunication networks, a given set of client nodes must be served by different sets of facilities—providing different services and having different capabilities—which must be located and dimensioned in the design phase. Network topology must be designed as well, by assigning clients to facilities and facilities to higher level entities, when necessary. We tackle a particular location problem, where two sets of facilities have to be located, and in which different devices can be installed at each site, providing different capacities at different costs. We optimize location and dimensioning of these facilities simultaneously. We introduce a compact formulation of that problem, we use discretization and Dantzig–Wolfe reformulation techniques to improve models, and we design an exact optimization algorithm. We test our approach on a set of instances derived from existing literature on facility location. © 2011 Wiley Periodicals, Inc. NETWORKS, 2012

Abstract The capacitated p‐median problem is the variation of the well‐known p‐median problem in which a demand is associated to each user, a capacity is associated to each candidate median, and the total demand of the users associated to the same median must not exceed its capacity. We present a branch‐and‐price algorithm, that exploits column generation, heuristics and branch‐and‐bound to compute optimal solutions. We compare our branch‐and‐price algorithm with other methods proposed so far, and we present computational results both on test instances taken from the literature and on random instances with different values of the ratio between the number of medians and the number of users. © 2004 Wiley Periodicals, Inc. NETWORKS, Vol. 45(3), 125–142 2005

The scope and character of today's computing environments are progressively shifting from traditional, one-on-one client-server interaction to the new cooperative paradigm. It then becomes of primary importance to provide means of protecting the secrecy of the information, while guaranteeing its availability to legitimate clients. Operating online querying services securely on open networks is very difficult; therefore many enterprises outsource their data center operations to external application service providers. A promising direction toward prevention of unauthorized access to outsourced data is represented by encryption. However, data encryption is often supported for the sole purpose of protecting the data in storage while allowing access to plaintext values by the server, which decrypts data for query execution. In this paper, we present a simple yet robust single-server solution for remote querying of encrypted databases on external servers. Our approach is based on the use of indexing information attached to the encrypted database, which can be used by the server to select the data to be returned in response to a query without the need of accessing the plaintext database content. Our indexes balance the trade-off between efficiency requirements in query execution and protection requirements due to possible inference attacks exploiting indexing information. We investigate quantitative measures to model inference exposure and provide some related experimental results.