This work presents a comprehensive evaluation of neural network graph compilers across heterogeneous hardware platforms, addressing the critical gap between theoretical optimization techniques and practical deployment scenarios. We demonstrate how vendor-specific optimizations can invalidate relative performance comparisons between architectural archetypes, with performance advantages sometimes completely reversing after compilation. Our systematic analysis reveals that graph compilers exhibit performance patterns highly dependent on both neural architecture and batch sizes. Through fine-grained block-level experimentation, we establish that vendor-specific compilers can leverage repeated patterns in simple architectures, yielding disproportionate throughput gains as model depth increases. We introduce novel metrics to quantify a compiler's ability to mitigate performance friction as batch size increases. Our methodology bridges the gap between academic research and practical deployment by incorporating compiler effects throughout the research process, providing actionable insights for practitioners navigating complex optimization landscapes across heterogeneous hardware environments.

The increasing complexity of IoT applications and the continuous growth in data generated by connected devices have led to significant challenges in managing resources and meeting performance requirements in computing continuum architectures. Traditional cloud solutions struggle to handle the dynamic nature of these environments, where both infrastructure demands and data analytics requirements can fluctuate rapidly. As a result, there is a need for more adaptable and intelligent resource management solutions that can respond to these changes in real-time. This paper introduces a framework based on multi-dimensional elasticity, which enables the adaptive management of both infrastructure resources and data analytics requirements. The framework leverages an orchestrator capable of dynamically adjusting architecture resources such as CPU, memory, or bandwidth and modulating data analytics requirements, including coverage, sample, and freshness. The framework has been evaluated, demonstrating the impact of varying data analytics requirements on system performance and the orchestrator's effectiveness in maintaining a balanced and optimized system, ensuring efficient operation across edge and head nodes.

Hierarchical federated learning (HFL) designs introduce intermediate aggregator nodes between clients and the global federated learning server in order to reduce communication costs and distribute server load. One side effect is that machine learning model replication at scale comes “for free” as part of the HFL process: model replicas are hosted at the client end, intermediate nodes, and the global server level and are readily available for serving inference requests. This creates opportunities for efficient model serving but simultaneously couples the training and serving processes and calls for their joint orchestration. This is particularly important for continual learning, where serving a model while (re)training it periodically, upon specific triggers, or continuously, takes place over shared infrastructure spanning the computing continuum. Consequently, training and inference workloads can interfere with detrimental effects on performance. To address this issue, we propose an inference load-aware HFL orchestration scheme, which makes informed decisions on HFL configuration, considering knowledge about inference workloads and the respective processing capacity. Applying our scheme to a continual learning use case in the transportation domain, we demonstrate that by optimizing aggregator node placement and device-aggregator association, significant inference latency savings can be achieved while communication costs are drastically reduced compared to flat centralized federated learning.

Fog computing enables the execution of IoT applications on compute nodes which reside both in the cloud and at the edge of the network. To achieve this, most fog computing systems route the IoT data on a path which starts at the data source, and goes through various edge and cloud nodes. Each node on this path may accept the data if there are available resources to process this data locally. Otherwise, the data is forwarded to the next node on path. Notably, when the data is forwarded (rather than accepted), the communication latency increases by the delay to reach the next node. To avoid this, we propose a routing mechanism which maintains a history of all nodes that have accepted data of each context in the past. By processing this history, our mechanism sends the data directly to the closest node that tends to accept data of the same context. This lowers the forwarding by nodes on path, and can reduce the communication latency. We evaluate this approach using both prototype- and simulation-based experiments which show reduced communication latency (by up to 23 percent) and lower number of hops traveled (by up to 73 percent), compared to a state-of-the-art method.

Serverless computing has become a popular part of the cloud computing model, thanks to abstracting away infrastructure management and enabling developers to write functions that auto-scale in a polyglot environment, while only paying for the used compute time. While this model is ideal for handling unpredictable and bursty workloads, cold-start latencies of hundreds of milliseconds or more still hinder its support for latency-critical IoT services, and may cancel the latency benefits that come with proximity, when serverless functions are deployed at the edge. Moreover, CPU power and memory limitations which often characterize edge hosts drive latencies even higher. The root of the problem lies in the de facto runtime environments for serverless functions, namely container technologies such as Docker. A radical approach is thus to replace them with a more light-weight alternative. For this purpose, we examine WebAssembly's suitability for use as a serverless container runtime, with a focus on edge computing settings, and present the design and implementation of a WebAssembly-based runtime environment for serverless edge computing. WOW, our prototype for WebAssembly execution in Apache QpenWhisk, reduces cold-start latency by up to 99.5%, can improve on memory consumption by more than 5×, and increases function execution throughput by up to 4.2× on low-end edge computing equipment compared to the standard Docker-based container runtime for various serverless workloads.

Complex distributed systems such as the ones induced by Internet of Things (IoT) deployments, are expected to operate in compliance to their requirements. This can be checked by inspecting events flowing throughout the system, typically originating from end-devices and reflecting arbitrary actions, changes in state or sensing. Such events typically reflect the behavior of the overall IoT system – they may indicate executions which satisfy or violate its requirements. This article presents a service-based software architecture and technical framework supporting runtime verification for widely deployed, volatile IoT systems. At the lowest level, systems we consider are comprised of resource-constrained devices connected over wide area networks generating events. In our approach, monitors are deployed on edge components, receiving events originating from end-devices or other edge nodes. Temporal logic properties expressing desired requirements are then evaluated on each edge monitor in a runtime fashion. The system exhibits decentralization since evaluation occurs locally on edge nodes, and verdicts possibly affecting satisfaction of properties on other edge nodes are propagated accordingly. This reduces dependence on cloud infrastructures for IoT data collection and centralized processing. We illustrate how specification and runtime verification can be achieved in practice on a characteristic case study of smart parking. Finally, we demonstrate the feasibility of our design over a testbed instantiation, whereupon we evaluate performance and capacity limits of different hardware classes under monitoring workloads of varying intensity using state-of-the-art LPWAN technology.

We present an orchestration scheme for Deep Neural Network (DNN) model serving, capable of computation distribution over the device-to-cloud continuum and low-latency inference. Our system allows automated layer-wise splitting of DNN structures and their adaptive distribution over compute hosts, providing an execution environment for collaborative inference. Model deployment and its self-adaptation at runtime are implemented by optimization algorithms supported in a plug-in manner. These follow service and infrastructure provider criteria and constraints, expressed via well-defined interfaces. Our framework can serve diverse neural architectures, including DNNs with early exits, with zero to minimal modifications.

Complex distributed systems such as the ones induced by Internet of Things (IoT) deployments, are expected to operate in compliance to their requirements. This can be checked by inspecting events flowing throughout the system, typically originating from end-devices and reflecting arbitrary actions, changes in state or sensing. Such events typically reflect the behavior of the overall IoT system – they may indicate executions which satisfy or violate its requirements.

One of the main challenges when it comes to deploying Network Slices is slicing the Radio Access Network (RAN). Indeed, managing RAN resources and sharing them among network slices is an increasingly difficult task, which needs to be properly designed. The goal is to improve network performance and introduce flexibility and greater utilization of network resources by accurately and dynamically provisioning the activated network slices with the appropriate amounts of resources to meet their diverse requirements. In this paper, we propose a data-driven RAN slicing mechanism based on a resource sharing algorithm running at theSlice Orchestrator(SO) level. This algorithm computes the necessary radio resources to be used by each deployed network slice. These resources are adjusted periodically based on current estimates of achievable throughput performance derived from channel quality information, and in particular from the Channel Quality Indicator (CQI) values of the users of each network slice retrieved from the RAN. CQI information is reported to base stations by the User Equipment (UE) following standard procedures, but extracting and frequently reporting it from base stations to the SO may result in significant communication overhead. To mitigate this overhead while maintaining at the SO level an accurate view of UE channel qualities, we propose a machine learning approach to infer the stability of UE channel conditions, as well as predictive schemes to reduce the CQI reporting intensity based on the inferred channel status. Through extensive simulations, we demonstrate the efficiency of our data-driven RAN slicing framework, which allows to meet the stringent requirements of two main classes of network slices in 5G, i.e., enhanced Mobile Broadband (eMBB) and Ultra-Reliable and Low-Latency Communication (URLLC).

Channel quality feedback is crucial for the operation of 4G and 5G radio networks, as it allows to control User Equipment (UE) connectivity, transmission scheduling, and the modulation and rate of the data transmitted over the wireless link. However, when such feedback is frequent and the number of UEs in a cell is large, the channel may be overloaded by signaling messages, resulting in lower throughput and data loss. optimizing this signaling process thus represents a key challenge. In this paper, we focus on Channel Quality Indicator (CQI) reports that are periodically sent from a UE to the base station, and propose mechanisms to optimize the reporting process with the aim of reducing signaling overhead and avoiding the associated channel overloads, particularly when channel conditions are stable. To this end, we apply machine learning mechanisms to predict channel stability, which can be used to decide if the CQI of a UE is necessary to be reported, and in turn to control the reporting frequency. We study two machine learning models for this purpose, namely Support Vector Machines (SVM) and Neural Networks (NN). Simulation results show that both provide a high prediction accuracy, with NN consistently outperforming SVM in our settings, especially as CQI reporting frequency reduces.

Multi-access Edge Computing (MEC) is one of the key enablers in 5G, where the objective is to bring computation very close to the end users. MEC, as defined by ETSI, introduces several services that can be exposed to MEC applications regarding the mobile users, such as the Radio Network Information Service (RNIS) and the Location Service, which provide low-level information on mobile users (e.g., Channel Quality Indicator - CQI), allowing the development of context-aware edge applications. In this paper, we address the challenging question of where to deploy a set of MEC applications on a federated edge infrastructure so as to meet the applications' requirements in terms of computing resources and latency, while ensuring that the MEC platform services required by each application are available at the selected edge locations. We formulate this service placement problem as an Integer Linear Program, which aims at balancing the computing load between available Mobile Edge Platforms (MEP), while respecting application latency and MEP service availability constraints. This problem is shown to be NP-hard. To solve it computationally efficiently, we propose an algorithm based on the Tabu-Search (TS) meta-heuristic. Via simulation, we demonstrate the efficiency of our scheme in balancing computational load among available MEPs and its ability to optimize service placement.

We present an architecture for the provision of video Content Delivery Network (CDN) functionality as a service over a multi-domain cloud. We introduce the concept of a CDN slice, that is, a CDN service instance which is created upon a content provider's request, is autonomously managed, and spans multiple, potentially heterogeneous, edge cloud infrastructures. Our design is tailored to a 5G mobile network context, building on its inherent programmability, management flexibility, and the availability of cloud resources at the mobile edge level, thus close to end users. We exploit Network Functions Virtualization (NFV) and Multi-access Edge Computing (MEC) technologies, proposing a system which is aligned with the recent NFV and MEC standards. To deliver a Quality-of-Experience (QoE) optimized video service, we derive empirical models of video QoE as a function of service workload, which, coupled with multi-level service monitoring, drive our slice resource allocation and elastic management mechanisms. These management schemes feature autonomic compute resource scaling, and on-the-fly transcoding to adapt video bit-rate to the current network conditions. Their effectiveness is demonstrated via testbed experiments.

Network slicing is one of the key components allowing to support the envisioned 5G services, which are organized in three different classes: Enhanced Mobile Broadband (eMBB), massive Machine Type Communication (mMTC), and Ultra-Reliable and Low-Latency Communication (URLLC). Network Slicing relies on the concept of Network Softwarization (Software Defined Networking - SDN and Network Functions Virtualization - NFV) to share a common infrastructure and build virtual instances (slices) of the network tailored to the needs of different 5G services. Although it is straightforward to slice and isolate computing and network resources for Core Network (CN) elements, isolating and slicing Radio Access Network (RAN) resources is still challenging. In this paper, we leverage a two-level MAC scheduling architecture and provide a resource sharing algorithm to compute and dynamically adjust the necessary radio resources to be used by each deployed network slice, covering eMBB and URLLC slices. Simulation results clearly indicate the ability of our solution to slice the RAN resources and satisfy the heterogeneous requirements of both types of network slices.

The Radio Network Information Service (RNIS) is one of the key services provided by a Multi-access Edge Computing Platform (MEP), as specified in the relevant ETSI MEC standards. It is responsible for interacting with the Radio Access Network (RAN), collecting RAN-level information about User Equipment (UE) and exposing it to mobile edge applications, which can in turn utilize it to dynamically adjust their behavior to optimally match the RAN conditions. Putting the provision of RNIS in the context of the emerging MEC-in-NFV environment, where the components and services of the MEC architecture, including the MEP itself, are integrated in an NFV environment and are delivered on top of a virtualized infrastructure, we present our standards-compliant RNIS implementation based on OpenAirInterface and study critical performance aspects for its provision as a virtual function. Since the RNIS design and operation follows the publish-subscribe model, we provide alternative implementations using different message brokering technologies (RabbitMQ and Apache Kafka), and compare their use and performance in an effort to evaluate their suitability for providing RNIS in an as-a-service manner.

The trend towards the cloudification of the 3GPP LTE mobile network architecture and the emergence of federated cloud infrastructures call for alternative service delivery strategies for improved user experience and efficient resource utilization. We propose Follow-Me Cloud (FMC), a design tailored to this environment, but with a broader applicability, which allows mobile users to always be connected via the optimal data anchor and mobility gateways, while cloud-based services follow them and are delivered via the optimal service point inside the cloud infrastructure. Follow-Me Cloud applies a Markov-decision-process-based algorithm for cost-effective performance-optimized service migration decisions, while two alternative schemes to ensure service continuity and disruption-free operation are proposed, based on either software defined networking technologies or the locator/identifier separation protocol. Numerical results from our analytic model for follow-me cloud, as well as testbed experiments with the two alternative follow-me cloud implementations we have developed, demonstrate quantitatively and qualitatively the advantages it can bring about.

Multi-access edge computing (MEC) comes with the promise of enabling low-latency applications and of reducing core network load by offloading traffic to edge service instances. Recent standardization efforts, among which the ETSI MEC, have brought about detailed architectures for the MEC. Leveraging the ETSI model, in this paper we first present a flexible, yet full-fledged, MEC architecture that is compliant with the standard specifications. We then use such architecture, along with the popular OpenAir interface (OAI), for the support of automotive services with very tight latency requirements. We focus in particular on the extended virtual sensing (EVS) services, which aim at enhancing the sensor measurements aboard vehicles with the data collected by the network infrastructure, and exploit this information to achieve better safety and improved passengers/driver comfort. For the sake of concreteness, we select the intersection control as an EVS service and present its design and implementation within the MEC platform. Experimental measurements obtained through our testbed show the excellent performance of the MEC EVS service against its equivalent cloud-based implementation, proving the need for MEC to support critical automotive services, as well as the benefits of the solution we designed.

Multi-access Edge Computing (MEC) is gaining momentum as it is considered as one of the enablers of 5G ultra-Reliable Low-Latency Communications (uRLLC) services. MEC deploys computation resources close to the end user, enabling to reduce drastically the end-to-end latency. ETSI has recently leveraged the MEC architecture to run all MEC entities, including MEC applications, as Virtual Network Functions (VNF) in a Network Functions Virtualization (NFV) environment. This evolution allows taking advantage of the mature architecture and the enabling tools of NFV, including the potential to apply a variety of service-tailored function placement algorithms. However, the latter need to be carefully designed in case of MEC applications such as uRLLC, where service access latency is critical. In this paper, we propose a novel placement scheme applicable to a MEC in NFV environment. In particular, we propose a formulation of the problem of VNF placement tailored to uRLLC as an optimization problem of two conflicting objectives, namely minimizing access latency and maximizing service availability. To deal with the complexity of the problem, we propose a Genetic Algorithm to solve it, which we compare with a CPLEX implementation of our model. Our numerical results show that our heuristic algorithm runs efficiently and produces solutions that approximate well the optimal, reducing latency and providing a highly-available service.

Envisioned low-latency services in 5G, like automated driving, will rely mainly on Multi-access Edge Computing (MEC) to reduce the distance, and hence latency, between users and the remote applications. MEC hosts will be deployed close to mobile base stations, constituting a highly distributed computing platform. However, user mobility may raise the need to migrate a MEC application among MEC hosts to ensure always connecting users to the optimal server, in terms of geographical proximity, Quality of Service (QoS), etc. However, service migration may introduce: (i) latency for users due to the downtime duration; (ii) cost for the network operator as it consumes bandwidth to migrate services. One solution could be the use of service replication, which pro-actively replicates the service to avoid service migration and ensure low latency access. Service replication induces cost in terms of storage, though, requiring a careful study on the number of service to replicate and distribute in MEC. In this paper, we propose to compare service migration and service replication via an analytical model. The proposed model captures the relation between user mobility and service duration on service replication as well as service migration costs. The obtained results allow to propose recommendations between using service migration or service replication according to user mobility and the number of replicates to use for two types of service.

The future 5G transport networks are envisioned to support a variety of vertical services through network slicing and efficient orchestration over multiple administrative domains. In this paper, we propose an orchestrator architecture to support vertical services to meet their diverse resource and service requirements. We then present a system model for resource orchestration of transport networks as well as low-complexity algorithms that aim at minimizing service deployment cost and/or service latency. Importantly, the proposed model can work with any level of abstractions exposed by the underlying network or the federated domains depending on their representation of resources.

We address the fundamental tradeoff between deployment cost and service availability in the context of on-demand content delivery service provision over a telecom operator's network functions virtualization infrastructure. In particular, given a specific set of preferences and constraints with respect to deployment cost, availability and computing resource capacity, we provide polynomial-time heuristics for the problem of jointly deriving an appropriate assignment of computing resources to a set of virtual instances and the placement of the latter in a subset of the available physical hosts. We capture the conflicting criteria of service availability and deployment cost by proposing a multi-objective optimization problem formulation. Our algorithms are experimentally shown to outperform state-of-the-art solutions in terms of both execution time and optimality, while providing the system operator with the necessary flexibility to balance between conflicting objectives and reflect the relevant preferences of the customer in the produced solutions.

We focus on the problem of optimal compute resource allocation and placement for the provision of a virtualized Content Delivery Network (CDN) service over a telecom operator's Network Functions Virtualization (NFV) infrastructure. Starting from a Quality of Experience (QoE)-driven decision on the necessary amount of CPU resources to allocate to satisfy a virtual CDN deployment request with QoE guarantees, we address the problem of distributing these resources to virtual machines and placing the latter to physical hosts, optimizing for the conflicting objectives of management cost and service availability, while respecting physical capacity, availability and cost constraints. We present a multi-objective optimization problem formulation, and provide efficient algorithms to solve it by relaxing some of the original problem's assumptions. Numerical results demonstrate how our solutions address the trade-off between service availability and cost, and show the benefits of our approach compared with resource placement algorithms which do not take this trade-off into account.

We focus on QoE-optimized video delivery in a wireless heterogeneous network setting, where users equipped with multi-interface devices access a Dynamic Adaptive Streaming over HTTP (DASH) video service. We provide an Integer Linear Programming (ILP) formulation for the problem of optimal joint video quality and network interface selection and a heuristic algorithm to solve it, shown via simulation and testbed experiments to achieve near-optimal performance in terms of Quality of Experience (QoE), with reduced execution time. We further design a video delivery architecture based on the emerging Mobile Edge Computing (MEC) standard, where our video quality and network selection functionality is executed as a MEC application. Notably, our architecture transparently operates with standard DASH clients.

We present the design and implementation of a content-delivery-network-as-a-service (CDNaaS) architecture, which allows a telecom operator to open up its cloud infrastructure for content providers to deploy virtual content delivery network (CDN) instances on demand, at regions where the operator has presence. Using northbound REST APIs, content providers can express performance requirements and demand specifications, which are translated to an appropriate service placement on the underlying cloud substrate. Our architecture is extensible, supporting various different CDN flavors, and, in turn, different schemes for cloud resource allocation and management. In order to decide on the latter in an optimal manner from an infrastructure cost and a service quality perspective, knowledge of the performance capabilities of the underlying technologies, and compute resources is critical. Therefore, to gain insight which can be applied to the design of such mechanisms, but also with further implications on service pricing and SLA design, we carry out a measurement campaign to evaluate the capabilities of key enabling technologies for CDNaaS provision. In particular, we focus on virtualization and containerization technologies for implementing virtual CDN functions to deliver a generic HTTP service, as well as an HTTP video streaming one, empirically capturing the relationship between performance and service workload, both from a system operator and a user-centric viewpoint.

The volume of mobile multimedia traffic is fast-growing, challenging the radio and backhaul network infrastructure and calling for alternative content dissemination schemes. To improve user experience and reduce infrastructure load, we exploit implicit social relationships among users and take into account content popularity, proposing push-based prefetching mechanisms which take advantage of the caching and mobile ad hoc networking capabilities of user devices. We use Bloom Filters as summaries of user caches, and design mechanisms to estimate the social distance between users and the popularity of content items, which drive our algorithms. Our simulation-based evaluation shows that our scheme brings caching performance improvements in an order of 10% in terms of absolute cache hit ratio in most of the cases studied, and from 3% to 82% in terms of normalized cache hit ratio gain.

To enhance video streaming experience for mobile users, we propose an approach towards Quality-of-Experience (QoE) aware on-the-fly transcoding. The proposed approach relies on the concept of Mobile Edge Computing (MEC) as a key enabler in enhancing service quality. Our scheme involves an autonomic creation of a transcoding service as a Virtual Network Function (VNF) and ensures dynamic rate switching of the streamed video to maintain the desirable quality. This edge-assistive transcoding and adaptive streaming results in reduced computational loads and reduced core network traffic. The proposed solution represents a complete miniature content delivery network infrastructure on the edge, ensuring reduced latency and better quality of experience.

The switch in video delivery from traditional medium to Over-The-Top (OTT) based services has been a game changer for stakeholders of the media delivery ecosystem. It has changed the value chain enabling Content Delivery Networks and Content Providers to take the lion's share at the expense of Internet/Network Service Providers. Even if establishing collaboration between them naturally appears as the key to provide good Quality of Service to the end-users, they struggle in finding efficient and fair ways to do so. This paper directly tackles this problem and proposes a new model for content delivery actors to collaborate over a Virtualized Infrastructure, fairly balancing the revenue stream created. We list the main challenges and the new technical opportunities to solve them, among which the deployment of a distributed Network Function Virtualization (NFV) platform at the edge of the Internet Service Provider's network where a virtual Content Delivery Network (vCDN) is proposed to be deployed. Furthermore, we apply a game-theoretic analysis to study different ISP-CDN collaboration models and identify optimality conditions for our proposed CDN-as-a-Virtual-Network-Function approach.

We focus on the provision of Content-Delivery-Network-as-a- Service (CDNaaS) functionality for video distribution and, in particular, on how to appropriately decide on the amount of computing resources to allocate to a CDNaaS instance to satisfy Quality of Experience (QoE) and resource capacity constraints. Our work is put in the context of a telco cloud environment, which is open for content providers to request the dynamic deployment of a virtual CDN infrastructure on top of it, to cover a target user demand with a desired service quality at the regions where the telecom operator has presence. We perform extensive testbed experiments to quantify the dependence of user experience on the load of a virtualized video service, and apply our measurement results to drive a QoE-aware virtual CPU resource allocation algorithm. Using simulation, we show our algorithm to optimally address the quality-cost tradeoff, demonstrating the benefits of QoE-awareness.

Internet Service Providers are becoming more involved in the audiovisual content delivery chain. One manifestation of this trend is the emergence of telco CDNs, i.e., content delivery networks operated by telecom service providers. In this work, we make the case for opening the telco CDN infrastructure to content providers by means of network function virtualization (NFV) and cloud technologies. We design and implement a CDN-as-a-Service architecture, where content providers can lease CDN resources on demand at regions where the ISP has presence. Using open northbound RESTful APIs, content providers can express performance requirements and demand specifications, which can be translated to an appropriate service placement on the underlying cloud substrate. To gain insight which can be applied to the design of such service placement mechanisms, we evaluate the capabilities of key enabling virtualization technologies by extensive testbed experiments.

We focus on the problem of efficiently integrating wireless users in future Information-Centric Networks (ICN), where communication is based on publish-subscribe primitives. The current host-centric Internet paradigm is abandoned in favor of information-oriented, rendezvous-based communication, where multicast data delivery is the norm. However, Wi-Fi, the predominant means of local wireless connectivity today, but also 3G and 4G technologies, are known to suffer from poor multicast performance. Data destined to a broadcast or multicast address are typically transmitted at lower rates to increase reliability for clients with poor signal conditions, causing unfavorable delays for high-rate users. One approach to this problem is to designate a subset of the clients as relays who re-broadcast packets for other clients at a higher rate. Given that different types of content have different performance requirements, we exploit content-awareness, inherent in our environment, to optimize for different criteria on a per-content basis. For this purpose, we provide a multi-objective optimization formulation for the problem of relay selection and rate assignment, which can capture the tradeoff among reliability, performance and energy cost.

The low cost and ease of installation of Wi-Fi equipment operating in unlicensed spectrum have made dense Wi-Fi deployments a reality in most modern urban areas. With the lack of many non-overlapping frequencies to operate on, interference among neighbor Wi-Fi cells can cause significant performance degradation. Here we study the problem of topology discovery in such dense deployments, which is necessary in order to combat interference. To this end, we apply a client-driven scheme, where client devices sense the spectrum and report overlapping cells. However, reporting entities cannot always be assumed trustworthy. We therefore study cases where reporters attack the discovery process by submitting fake information and propose simple countermeasures to tackle some attacker strategies. We show analytically and via simulations that, in realistic urban scenarios, our mechanisms are effective, even in the presence of a large number of attackers.

In densely populated cities, Wi-Fi networks-private or otherwise-are ubiquitous. We focus on the provision of citywide broadband communication capability to mobile users through private Wi-Fi networks that are in range but belong to others. We form a club that relies on indirect reciprocity: Members participate in the club and provide free Wi-Fi access to other members in order to enjoy the same benefit when they are away from their own Wi-Fi network. Our club scheme does not require registration with an authority and does not rely on centrally issued club identities: Members create their own identities (public-private key pairs) and receive signed digital receipts when they provide Wi-Fi service to other members. These receipts form a distributed receipt graph, parts of which are used as input to an indirect reciprocity algorithm that classifies club members according to their contribution. We show that our algorithm can sustain cooperation within the club and is robust to attacks by free-riders. We implement and evaluate our proposed club algorithms on commodity Wi-Fi routers and dual-mode cellular/Wi-Fi phones. Because we anticipate that Wi-Fi telephony will be a popular club application, we present and evaluate a secure and decentralized architecture for citywide voice (and multimedia) communications that is compatible with our club both from an architectural as well as an incentives perspective.

In recent years, we have witnessed a trend towards open wireless access, mainly driven by the low cost of IEEE 802.11-capable equipment and its operation in unlicensed spectrum. This trend has many faces; municipalities set up free Wi-Fi hotspots for Internet connectivity in public spaces, while Wi-Fi access is provided as an extra service to customers in other venues, such as restaurants or hotels. Also, in many metropolitan areas worldwide, community-initiated wireless mesh networks have emerged. Their members use inexpensive technologies to build multi-hop wireless networks and communicate autonomously. In this work, we document the structure and evolution of one of the largest community wireless mesh networks worldwide, the Athens Wireless Metropolitan Network (AWMN). We focus on how the network has grown in size, complexity and service offerings in the last few years. To be able to dynamically discover the structure of such a network, we have developed a suitable topology discovery methodology.

IEEE 802.11k is an extension of the IEEE 802.11 specification for radio resource measurements. In an IEEE 802.11k-enabled wireless LAN, an access point or other network element may request from a client or another access point to monitor and report the load of a channel. We call the latter a channel monitoring station. In this paper we propose a mechanism for a channel monitoring station to efficiently derive accurate values of channel load.We especially focus on optimizing the duration of channel monitoring and thus minimize the impact on applications. Note that such mechanisms are critical for the success of new sharing regimes such as Cognitive Radio and Open Spectrum Access.

Driven by their low cost and ease of deployment, as well as their operation in unlicensed spectrum bands, IEEE 802.11-based Wireless Local Area Networks (WLANs), also termed Wi-Fi, have been established as the de facto access technology for local area wireless connectivity. Especially in densely-populated urban areas, WLAN presence is ubiquitous. Residential WLAN owners, municipalities and venue owners, among others, set up wireless hotspots for private or public use. However, one can identify two important problems that have to be efficiently tackled. On the one hand, while Wi-Fi is present practically everywhere in modern metropolitan areas, access to roaming users is typically restricted. At the same time, the broadband Internet connections WLAN Access Points (APs) are attached to may have excess capacity, thus leaving resources underutilized. On the other hand, unplanned Wi-Fi deployment leads to significant interference problems among neighbor WLANs. In this work, we exploit our prior work on WLAN sharing communities to jointly tackle the above problems. The solution we propose is based on offering users QoS benefits as an incentive to perform spectrum sensing and supply interference reports to the WLAN APs they are connected to. We present the design of our mechanisms, discuss some of their properties and investigate their expected performance overhead, especially on delay-sensitive applications like VoIP.

The use of licensed spectrum for wireless communication is driven by the need to control interference between different operators. However, with this mode of regulation, spectrum utilization is far from efficient and the growth of wireless networks is hindered by the shortage of free frequency bands and the vast investments for the acquisition of a license. In view of this situation, we present an alternative evolution path for the unobstructed growth of wireless networks and the efficient use of spectrum. The proposed architecture is based on the use of unlicensed spectrum and the open access of users to all public networks without prior contracts with operators. We highlight and discuss the inherent technical challenges that must be tackled before the proposed solution can be realized. Special attention is paid to the inherent need for alternative interference mitigation strategies.

Abstract—Wireless Community Networks (WCNs) are wide-area wireless networks whose nodes are owned and managed by volunteers. We focus on the provision of Internet access to mobile users through WCN-controlled wireless LAN access points (APs). We rely on reciprocity: a person participates in the WCN and provides ‘free ’ Internet access to mobile users in order to enjoy the same benefit when mobile. Our reciprocity scheme is compatible with the distinctive structure of WCNs: it does not require registration with authorities, relying only on uncertified free identities (public-private key pairs). Users sign digital receipts when they consume service. The receipts form a receipt graph, which is used as input to a reciprocity algorithm that identifies contributing users using network flow techniques. Simulations show that this algorithm can sustain reciprocal cooperation. We have implemented our algorithm to run on common APs. I.

We have designed and implemented P2PWNC, a fully distributed, open to all, autonomous WLAN roaming scheme that can be used in a community to provide various multimedia communication services. Here we discuss various security related issues and the support provided by P2PWNC for several aspects of communication and roaming privacy. We also report on aspects of its performance, focusing on the performance penalty of security-related operations. The core P2PWNC scheme assumes that community members are selfish and do not trust each other and uses a secure incentive technique to encourage their contribution. It protects the real-world identities of community providers and clients by relying only on disposable opaque identifiers and does not rely on any authority to resolve disputes or control membership. It could easily complement cellular networks for low-mobility users in metropolitan areas where some wireless community networks provide wide coverage