Search

Over the years, wireless technologies have enabled innovation in the industry and leveraged social and economic changes. With the so-called Industry 4.0, future digital industries are moving towards the distributed organization of production, with connected goods, low energy processes, collaborative robots, and integrated manufacturing and logistics. 5G is expected to guarantee stringent Quality of Service (QoS) delivery and for that to happen, the proper employment of wireless Medium Access Control (MAC) protocols is essential for efficient and reliable wireless communication. In indoor scenarios, given its low cost and easy deployment, IEEE 802.11 networks are still the default access choice. Among such stringent and heterogeneous requirements, network slicing is envisioned as the architecture to carve out multiple virtual networks with significantly different performance, security, and traffic isolation characteristics from a common physical infrastructure. However, deciding how to efficiently allocate, control, and manage users and slices remains challenging. Besides, although Software-Defined Networking (SDN) has enabled new levels of innovation and automation for the creation of Resource Allocation (RA) mechanisms, latency-related metrics are often neglected. In this thesis, we analyze the evolution and programmability of wireless MAC protocols and present the questions that should be answered to determine the MAC programmability needed. With our lessons learned, we identify the challenges and opportunities for future-proof MAC designs. Next, we propose an airtime-based slice orchestration approach and RA modeling for IEEE 802.11 Radio Access Network (RAN). By orchestrating airtime portions at the IEEE 802.11 upper MAC layer, we show how our network slicing algorithm can enhance the QoS at the RAN. We present our airtime-based RA modeling for network slicing in IEEE 802.11 RAN and the limitations of performing such optimizations at runtime. To support 5G Mission-Critical Applications (MCAs), we propose a delay-aware approach for MAC management via airtime-based network slicing and traffic shaping, as well as user association using Multi-Criteria Decision Analysis (MCDA). Through experimentation in a real-world testbed, our approach maintains the queueing delay requirements of 5ms under varying traffic demands and for most of the experiments run. Last, we provide a complete Software-Defined-RAN interactive management approach using In-band Network Telemetry (INT) in IEEE 802.11 networks. With INT, the end-to-end flow dynamics, state of the wireless links, and per-hop reliability can be fully assessed.

To respond to innovation aspirations from the digital society, networks need to be significantly more flexible and adjustable. Such flexibility can be achieved through network softwarization, which makes network infrastructure fully programmable/adjustable. The two main pillars of network programmability are Network Function Virtualization (NFV) and Software Defined Networking (SDN), where NFV virtualizes network resources and enables the dynamic creation of Virtualized Network Functions (VNFs), and SDN programs the way those VNFs are connected to each other over the network. The 5G ecosystems, consisting of 5G New Radio, edge, transport, and core network, are robust software-based systems that rely on both SDN and NFV to enable network programmability. By applying the concept of virtualization (e.g., in the core and edge networks, and partially on the radio side), more and more resources are virtualized, resulting in an enormous heterogeneity across the network ecosystems. Thus, one of the significant challenges lies in agile management and orchestration of those resources that are diverse and distributed across the overall 5G ecosystem. The MANagement and Orchestration (MANO) role is to identify the needs of traffic and services running on the virtualized network infrastructure and dynamically respond to those needs by performing service reconfigurations, such as scaling, migration, and service teardown, which are known as MANO operations. Throughout this Ph.D. research, we have been focused on investigating service and resource management and orchestration in Multi-Access Edge Computing (MEC)-enhanced 5G networks to enable openness and programmability of 5G and beyond ecosystems. The main objective of this research is to leverage 5G, MEC, and Artificial Intelligence (AI)/Machine Learning (ML), to perform efficient and automated MANO operations across different technological/administrative edge domains, and to achieve the low-latency-aware VNF placement and seamless migration of programmable Network Services (NSs). One of the main components of this research is its applicability to the next-generation real-world networks and systems, as it has been driven by an active engagement with industry, which resulted in strong collaboration with several important partners from 5G, automotive, and transport & logistics industries. Moreover, the thesis contributed to the following European projects: H2020 5G-CARMEN, H2020 DAEMON, H2020 VITAL-5G, H2020 5G-BLUEPRINT, and H2020 FED4FIRE+, resulting in ve journal papers (published and submitted), 12 publications in conference proceedings (full, work-in-progress, and demo papers; published and submitted), and a book chapter, as a first author, as well as three journals and six conference papers, as a co-author (published and submitted).

Wireless networks are a very complex and dynamic environment that requires proper management to optimise its performance. With a tremendous increase in the number of user devices and their data traffic, wireless networks are becoming denser. Therefore, optimising mobility management becomes a crucial challenge. In heterogeneous wireless networks, the usage of multiple wireless interfaces is inefficient and suboptimal. Solutions require modifications to the user device limiting the use cases. Reactive approaches are too slow to react and solutions often reside on the user side, which creates suboptimal perdevice optimisations. To tackle these challenges we use SDN. SDN monitors the network by collecting multiple metrics on the network side and creates a global overview of the network. This overview can be used in SON algorithms to proactively trigger global optimisations. We create and use different SDN frameworks for different wireless technologies and create SON algorithms on top to optimise mobility. In heterogeneous wireless networks, we introduce ORCHESTRA, an SDN framework to manage different devices and do packetlevel dynamic and intelligent handovers, load balancing and replication. On top of ORCHESTRA we present a novel MIQP load-balancing optimization problem formulation for mobility management which can double the network-wide throughput across different scenarios. To avoid user device modifications, we introduce the HuMOR framework for handover management in WiFi. On top of HuMOR we introduce ABRAHAM a handover algorithm that uses multiple metrics to predict the future state of the network and optimize the AP load. We show that ABRAHAM can achieve a 139% throughput improvement over the IEEE 802.11 handover algorithm. To demonstrate that our WiFi solution is capable to address different wireless contexts, we use the same principles in mobile networks. On top of an OpenRAN compliant framework dRAX, we introduce MOLA_ADNA. This adaptive and QoS aware handover algorithm takes multiple metrics into account to do network level optimization of the cell load. We show that the mean throughput with MOLA-ADNA was higher by 25% compared to A3. Finally, proper network element configuration is also crucial for network performance and mobility management. We, therefore, create the ALPACA algorithm on top of dRAX to optimise the PCI value of cells in mobile networks. ALPACA adapts to dynamic network topology changes and continuously optimizes the network. The results show that ALPACA can resolve all collisions and confusion for up to 1000 cells and minimize the effects of inevitable modulo PCI issues.

The last two decades brought a phenomenal increase in communication devices, mobile and stationary. The overall number of connected devices went past 18 billion with many available technologies. Two wireless technologies dominate the market space: LTE/5G and Wi-Fi. Other new wireless technologies also rose in popularity. This situation gave rise to new applications that require high bandwidth and low latency. However, technologies share use cases, devices can not fully use their technologies, and technologies can interfere with each other. Avoiding interference and enabling technology cooperation are major cornerstones of improving user experience. Nevertheless, cooperation can not overcome all obstacles. Estimation and modeling of performance within certain environments are necessary. Current solutions for integrated technology management are limited. They focus on specific technologies, specific use cases, or have limited management capability. Performance modeling is more common but focuses on interference from other communication technologies, not interference from generic electronic devices. The increasing use of electrical devices multiplies the problem and affects many devices and technologies. This dissertation provides three significant contributions to address these challenges. The first contribution explores and models IEEE 802.11 systems' performance when an interfering source is present that is not a communication technology. We first explore the performance of IEEE 802.11 in a challenging environment via a wireless mesh network and further in a controlled setup and simulation. We provide two models. The first is based on base performance when no interference is present and is computationally fast. The second is an analytical model that models the entire system's behavior but is computationally expensive. The second contribution consists of the ORCHESTRA framework, enabling inter-technology management seamlessly to the user and operator. This framework offers interference mitigation by using multiple technologies. It uses technology abstraction through a virtual layer and advanced packet-level functionalities, such as handovers, load balancing, and duplication. A central controller maintains a global view of the network and makes intelligent decisions to improve performance. As a third contribution, we present a load balancing solution that normalizes latency for links with different latency properties. Different technologies exhibit different latency properties that cause problems when using packet-level load balancing. We provide a machine learning based normalization method that smooths and reduces latency on a flow. Instead of sending out bursts of packets after reordering, packets are sent with a short time in between to avoid burst behavior.

The Internet of Things (IoT) paradigm is the shift towards a world where all things are connected to the Internet. While nowadays IoT has an impact on every aspect of society, it is being applied particularly efficiently to further revolutionize the automation and control of traditional manufacturing and industrial processes, leading to the term Industrial Internet of Things (IIoT). To fulfill the high-end requirements of IIoT applications, interconnecting all sensing and actuating devices was initially typically done through wiring. While the reliability advantage of wiring is obvious, it is costly and can be impractical in hard to reach locations or mobile machinery. Therefore, the transition to wireless communication seemed unavoidable and is becoming more and more ubiquitous. However, for this transition to be successful, wireless communication should show wire-like reliability in harsh industrial environments that suffer from external interference and multi-path fading effects that may easily disrupt the wireless signal. Additionally, to avoid having to frequently replace batteries in inconvenient places, the wireless devices should be able to run on limited battery capacity for years. Therefore, while required to be highly reliable, they should also exhibit low-power operations. A relatively recent wireless technique that has proven to be successful in fulfilling these requirement, is IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH), that combines frequency diversity with strict time-synchronization, achieving wired-like reliability of more than 99.999% while having ultra-low power consumption. ​ ​ In this PhD thesis, I study the TSCH Medium Access Control (MAC) layer and how it can be further improved to deploy it successfully in industrial networks. More specifically, I focus on 3 research questions related to the energy consumption, latency and reliability of TSCH networks. First, I investigate how the TSCH power consumption can be precisely characterized. Second, I aim at minimizing the communication delay of recurrent monitoring data that is typical for IIoT applications. Finally, I aim at further improving the industrial network’s overall reliability by introducing a technique to allows multiple physical (PHY) layers simultaneously in a single TSCH network. As such, each device is able to adapt its PHY layer to the link’s propagation characteristics. Additionally, a heuristic is also proposed that helps a device in making an appropriate parent and PHY selection in multi-PHY TSCH networks. In summary, this PhD thesis targets an energy-efficient, responsive and reliable TSCH network, thereby contributing to a wireless network deployment that is ready for the IIoT.

Wireless communication technologies became a part of our modern society. Every year the number of wireless devices and wireless technologies increases. Cisco expects that around 25.4 and 42.6 billion wireless devices will be connected to the Internet in 2022. This growth introduces some major challenges. One of these challenges is to use the wireless spectrum, used by all of these wireless devices, more efficient, especially within the radio bands itself. To meet the demand of more wireless devices and higher throughput, new techniques are necessary to optimise the use of the wireless spectrum. Based on literature, it was expected that collaboration between neighboring wireless networks (from all kind of technologies) can improve the efficiency of the use of the wireless spectrum. Increasing spectrum efficiency can be accomplished in two ways: (i) the improvement of physical transmission, (ii) the use Artificial Intelligence (AI) to improve the decisions made by the wireless nodes. The improvement of the physical transmissions has a direct effect on the efficiency of the use of the spectrum. It is clear that the more efficient data can be transmitted, to improve the bits per Hertz, the less spectrum will be used for the same data. AI, on the other hand, gives us the opportunity make smarter decisions based on the physical limitations of the wireless system and behavior of the environment. The use of AI can also enable the possibility to start and maintain collaboration with other neighboring technologies to improve the efficiency of the wireless spectrum. This dissertation focuses on the contributions made for the AI decision engine for wireless network technologies. Within the context of this dissertation we focus on the use of AI to improve the decisions made by the wireless nodes. This dissertation provides multiple improvements to enable collaboration for wireless networks, which will lead to a more efficient use of the wireless spectrum. First, we describe a decision-making framework designed to enable AI-enabled algorithms within wireless radio stacks. All other algorithms described in this dissertation are implemented within the framework. Secondly, we present a spectrum prediction algorithm. This prediction algorithm is able to predict the behavior of neighboring wireless networks, even if insufficient information is available. This ability provides us to select better transmission moments. Finally, we introduce the policy-based flow selection algorithm. This algorithm is able to collaborate to improve the Quality of Service and optimize the spectrum footprint.

Current Industrial Internet of Things (IoT) requires high-reliability, guaranteed performance and low-power consumption in order to allow mission-critical infrastructures monitoring and control in harsh environments with a battery lifetime of years. 6TiSCH is a promising Industrial IoT technology that aims to provide these requirements by combining the TSCH mode of the IEEE 802.15.4e standard with an IPv6-enabled upper stack that connects the IoT network to the Internet. Since 6TiSCH is based on TSCH, it has the capability of allocating a specific amount of bandwidth per node in a quasi-deterministic manner and at the same time, is able to minimize the radio duty cycle to reduce energy consumption. Additionally, the channel hopping technique addresses unreliability issues caused by factors such as multi-path fading and narrow-band external interference. However Industrial IoT networks and also 6TiSCH networks, are still facing major challenges. First, the current exponential growth of the IoT is also being experienced in the Industrial IoT. This calls for more scalable solutions that can cope with current and future demand. Second, Industrial IoT requires also to be flexible and programmable in order to be further tailored to the actual dynamic industrial automation needs. This is of most importance since Industrial IoT networks are expected to not only cope with a dynamic external environment that alters the available resources over time, but also to cope with variable demands and QoS policies that have different optimization goals over the time. Consequently, in order to achieve an effective and end-to-end QoS compliance, this programmability should be present in all the industrial network domains, including the IoT network. These challenges currently remain as open questions. On one hand, intradomain scalability largely depends on how network resources are shared and distributed. Although TSCH itself does not specify how these resources have to be scheduled, 6TiSCH provides the upper layers with minimal distributed scheduling capabilities that ensure communication between nodes. However, as we will later study in this book, limitations may appear when 6TiSCH networks scale up. On the other hand, flexibility and programmability are currently absent in 6TiSCH networks. In 6TiSCH, distributed protocols make static decisions over the routing (RPL) and scheduling (6P), and configuring the network dynamically is problematic. Alternatively, the proposed centralized SDN-based approaches provide operators with the required flexibility, but seem to be prone to high control overhead, and inherently, to be low scalable. On top of this, current Industrial IoT solutions are usually isolated in network silos, which hinders a complete end-to-end control in a multi-domain hierarchical architecture, which is typically required to scale the network up with QoS guarantees. This PhD book addresses these research questions. It first studies intradomain scalability in 6TiSCH networks, from a theoretical point of view and through extensive simulations. Then, it identifies the existing scalability problems and proposes a new scheduling mechanism in order to try to overcome them. Secondly, it proposes an innovative technique to flexibly control Industrial IoT networks (in our case 6TiSCH) in an efficient but yet scalable manner. We analyze our new approach in depth, discussing its possibilities and comparing it with other alternative approaches. Subsequently, we integrate it in a widely used open-source SDN framework to support multi-domain scalable end-to-end control in Industrial IoT networks. This means to allow a controller to be able to globally and jointly control,in a hierarchical manner, the wired and wireless network segments involved in a real Industrial IoT network. Finally, this PhD book also includes two complementary studies on 6TiSCH. First, we study the open-source simulator for 6TiSCH that allows benchmarking and fast-prototyping. The simulator is validated with current state-of-the-art hardware and software and is proven to be a useful tool to not only study scalability in 6TiSCH networks, but also the performance of different networks scenarios (topologies, traffic patterns, etc.), scheduling functions and routing algorithms. Secondly, in order to further discuss the potential and possibilities of Industrial IoT networks, this PhD book also includes a report on a cycling use case that leverages 6TiSCH to monitor athletes using a long-range multihop highly dynamic network. This use case, which may open new interesting research paths, shows that applications that intuitively would not have been the target for 6TiSCH at first sight, can also take advantage from 6TiSCH industrial performance.

The next generation of Internet services, e.g., self-driving cars, augmented reality and cloud robotics, requires ultra-low latency wireless communications, produces vast quantities of data and requires the speedy deployment of real-time collaborations between a wide variety of devices. While traditionally cloud services are hosted on infrastructure that is located within a single data center, these novel services have throughput and response time requirements that dictate that at least some computational tasks are executed near the location of the end user. Centralized and geo-distributed cloud environments are worlds apart. Compared to centralized clouds, geo-distributed cloud environments are much more heterogeneous. These environments incorporate both infrastructure in data centers and infrastructure at the network edge with very limited capabilities and that is much more failure-prone. This spread on capability; reliability; connectivity; and proximity to the end-user, severely complicates the management. This thesis investigates the challenges related to the orchestration of network services across heterogeneous cloud environments and proposes novel management approaches that address these challenges. First, we investigate how to effectively replicate data across storage nodes in these environments. We approach this problem as a runtime revenue problem, that considers both Service Level Agreements (SLAs) regarding durability and the cloud characteristics. This approach builds on a dynamic availability model that considers both the impact of failure distribution and recovery times on data loss. Second, we investigate how to protect stateless network services against a combination of node and link failure in these environments. We approach the problem of placing applications while guaranteeing a minimum availability for each application and minimizing the placement cost as a resource allocation problem. To deal with the scarcity of resources at the edge and the reliability spread, our availability-aware approach introduces protections only where they are needed. Third, we investigate how to orchestrate network services in a Network Functions Virtualization (NFV) environment. We propose orchestration algorithms that can improve the acceptance ratio and placement quality through coordination of the service composition and embedding. Not only do we develop the required orchestration algorithms for an existing service model that can generate Virtual Network Function (VNF)-Forwarding Graphs (FGs) with a tree topology, we also develop a novel service model with improved applicability and develop the required orchestration algorithms.

The Internet of Things (IoT) introduces a novel dimension to the world of information and communication technology where connectivity is available anytime, anywhere for anything. To make this into reality, a large number of battery-powered smart things, such as sensors and actuators, need to be connected to the Internet in an energy efficient manner. The newly released IEEE 802.11ah Wi-Fi standard, marketed as Wi-Fi HaLow, is considered as a very promising technology for the IoT. One of the new features of IEEE 802.11ah, named Restricted Access Window (RAW), aims to increase efficiency in face of a large scale of densely deployed, energy constrained stations. It divides stations into groups, limiting simultaneous channel access to one group, therefore reducing the collision probability and increase scalability in IoT networks. The IEEE 802.11ah standard, however, does not specify how to configure the actual RAW grouping parameters. Existing research has shown that the optimal RAW configuration depends on a variety of network-related parameters, such as the number of stations, traffic patterns, and network load. Incorrect configuration severely impacts throughput, latency and energy efficiency. Moreover, network conditions may change over time due to a variety of reasons. Therefore, this thesis aims to dynamically optimize RAW configurations in real time to adapt to the current network conditions. In order to accomplish this goal, there should be five approaches involved, including parameters evaluation, optimization, modeling, network condition estimation and implementation. (1) With respect to evaluation, an in-depth analysis needs to be conducted, in order to gain insight into the impact of network-related parameters on optimal RAW configuration. (2) Optimization takes a RAW performance model and network conditions as input, and generates RAW configurations as output, according to the pursued performance metrics. (3) Therefore, a model is required to be able to predict RAW performance for a given set of parameters under specific network and traffic conditions. (4) Moreover, it is indispensible to estimate network conditions since they are unknown and even dynamic. (5) Last but not least, as hardware for IEEE 802.11ah is not on the market yet, a realistic simulator is vital for accomplishing the parameters evaluation, and evaluating the effectiveness of the other three approaches. As IEEE 802.11ah targets large scale IoT networks, this thesis focuses on scenarios where each sensor transmits packets with a certain (predictable) frequency and may change the transmission frequency over time. Firstly, the implementation of IEEE 802.11ah is detailed, along with experimental results to validate it. The simulator consists of physical layer models an implementation of the MAC features, including not only RAW but also fast association, an energy state model, adaptive modulation and coding schemes (MCSs), and Traffic Indication Map (TIM) segmentation. Subsequently, the RAW performance is evaluated, using the implemented IEEE 802.11ah simulator. The simulation shows that, with appropriate grouping, the RAW mechanism substantially improves throughput, latency and energy efficiency. Furthermore, the results suggest that the optimal grouping strategy depends on many parameters, and intelligent RAW group adaptation is necessary to maximize performance under dynamic conditions. The results provide a major leap towards such a strategy. Moreover, the Traffic-Aware RAW Optimization Algorithm (TAROA) is proposed to adapt the RAW parameters in real time based on the current traffic conditions. TAROA introduces a traffic estimation method to predict the packet transmission interval of each station only based on packet transmission information obtained by the AP during the last beacon interval. By using the simulation results under saturated state as an alternative to the RAW model, TAROA derives the optimal number of stations to assign to a group according to the estimated traffic conditions, in order to maximize the throughput. In addition, an improved version of traffic estimation is proposed and integrated into an enhanced version of TAROA, referred to as Enhanced Traffic-Aware RAW Optimization Algorithm (E-TAROA). Both TAROA and E-TAROA supports homogeneous stations, i.e., all stations use the same MCS and packet size. A further step is made by applying surrogate modelling in this research. A surrogate model is an efficient mathematical representation of a black box system, it is based on supervised learning (e.g., Kriging, or neural networks), and can be accurately trained with very a few labeled sample data points. This research integrates the surrogate modeling toolbox into the implemented IEEE 802.11ah simulator, and trains models for estimating performance under a wide range of network and traffic conditions. Based on the trained models, the Model-Based RAW Optimization Algorithm (MoROA) is proposed. MoROA inherits the traffic estimation method of TAROA, using the trained model to determine the optimal RAW configuration in real time through multi-objective optimization. As a single trained model only supports homogeneous stations, MoROA supports heterogeneous networks in terms of MCS and packet size, by introducing multiple RAW groups and assigning all homogeneous stations into a single RAW group. Finally, an advanced surrogate model is presented that can predict performance of heterogeneous stations. As heterogeneous networks have more parameters leading to an enormous design space, the training methodology is well designed, in order to speed up the training process and maintains relative high model accuracy. In summary, the research develops solutions for real-time RAW optimization to support large scale IoT networks. An open-source IEEE 802.11ah simulator, a patent, multiple journal and conference papers have been produced as a result.

Tijdens de afgelopen decennia is er een enorme toename geweest in het gebruik en de beschikbaarheid van allerhande draadloze netwerken en toestellen. De massale adoptie van mobiele apparaten en toestellen gerelateerd aan het Internet der dingen liggen aan de oorzaak van deze toename. Daarnaast, zijn de eisen en verwachtingen van gebruikers omtrent connectiviteit, bandbreedte, kwaliteit, en functionaliteit, nog nooit zo hoog geweest. Deze evoluties hebben geleid tot een complexe en heterogene omgeving waar verschillende toestellen, applicaties, en technologieën naast elkaar bestaan op identieke of overlappende fysieke locaties. Helaas worden deze netwerken typisch op een statische manier aangestuurd aangezien er geen coördinatie bestaat tussen verschillende communicatietechnologieën en toestellen. Ondanks dat moderne toestellen zijn uitgerust met verschillende netwerkinterfaces, opteren ze typisch voor het statisch selecteren van één van de beschikbare technologieën of toegangspunten op basis van op voorhand gedefinieerde prioriteiten. Het gevolg is dat beslissingen worden overgelaten aan de applicaties of aan de gebruiker en het onmogelijk is om automatisch te reageren op de onvermijdelijke dynamische onderbrekingen of veranderingen binnen de draadloze omgeving. Ten slotte, beconcurreren verschillende technologieën elkaar vaak binnen dezelfde frequenties. In dit proefschrift proberen we, in tegenstelling tot de bestaande oplossingen, de opgenoemde uitdagingen op te lossen in een transparante en fundamentele manier. De belangrijkste leidraad hierbij is dat een gebruiker zich enkel bewust zou mogen zijn van het feit dat zijn toestel verbonden is met het internet, terwijl het netwerk alle onderliggende beslissingen neemt en voorziet in de nodige allocaties. Daarom stellen we de ORCHESTRA-omgeving voor dat naadloos beheer over verschillende technologieën heen mogelijk maakt. De omgeving bestaat uit twee componenten: een gecentraliseerde controller die zorgt voor een globaal netwerkoverzicht en een virtuele MAC laag die zorgt voor een enkel verbindingspunt voor de hogere lagen, door de onderliggende netwerktechnologieën op een transparante manier samen te voegen. Dit maakt handovers tussen de verschillende technologieën, het verdelen van de netwerkbelasting, en de duplicatie van datastromen mogelijk op het niveau van individuele datapakketten. Bovenop de ORCHESTRA-oplossing, is er nood aan intelligentie die de functionaliteiten van de omgeving gebruikt om het netwerk optimaal te gaan configureren. Daarom stellen we een aantal algoritmen voor die de hoeveelheid verkeer binnen een netwerk significant kunnen verhogen. Ten slotte, onderzoeken we of het mogelijk is om dataverkeerspatronen te detecteren in het radiospectrum. Deze informatie kan gebruikt worden door algoritmen om het netwerk verder te optimaliseren, in het bijzonder in situaties met verschillende overlappende netwerken.