Time-Sensitive Networking (TSN) is a toolbox of technologies that enable deterministic communication over Ethernet. A key area has been TSN’s time-aware traffic shaping (TAS), which supports stringent end-to-end latency and reliability requirements. Configuration of TAS requires the computation of a network-wide traffic schedule, which is particularly challenging with integrated wireless networks (e.g., 5G, Wi-Fi) due to the stochastic nature of wireless links. This paper introduces a novel method for configuring TAS, focusing on cyclic traffic patterns and jitter of wireless links. We formulate a linear program that computes a network-wide time-aware schedule, robust to wireless performance uncertainties. The given method enables robust scheduling of multiple TSN frames per transmission window using a tunable robustness parameter (Γ). To reduce computational complexity, we also propose a sequential batch-scheduling heuristic that runs in polynomial time. Our approach is evaluated by using different network topologies and wireless link characteristics, demonstrating that the heuristic can schedule 90% of 6500 requested TSN streams in a large topology.

Real-time traffic monitoring is critical for network operators to ensure performance, security, and visibility—especially as encryption becomes the norm. AI and ML have emerged as powerful tools to create deeper insights from network traffic, but collecting the fine-grained features needed at terabit speeds remains a major bottleneck. We introduce Direct Feature Access (DFA): a high-speed telemetry system that extracts flow features at line rate using P4-programmable data planes, and delivers them directly to GPUs via RDMA and GPUDirect —completely bypassing the ML server’s CPU. DFA enables feature enrichment and immediate inference on GPUs, eliminating traditional control plane bottlenecks and dramatically reducing latency. We implement DFA on Intel Tofino switches and NVIDIA A100 GPUs, achieving extraction and delivery of over 31 million feature vectors per second—supporting 524,000 flows within sub-20 ms monitoring periods—on a single port. DFA unlocks scalable, real-time, ML-driven traffic analysis at terabit speeds, pushing the frontier of what is possible for next-generation network monitoring.

Power efficiency is a growing concern in softwarized 5G gNBs due to increasing energy costs and carbon emissions. While SmartNICs offer low-power acceleration for packet processing, they struggle with complex operations, often requiring offloading to servers, which raises power usage. This study evaluates the performance and power consumption when dividing 5G gNB data plane tasks between SmartNICs and a DPDK-enabled host, comparing flow-based and function-based task allocation methods. We introduce an adaptive CPU power management strategy that adjusts CPU power states based on traffic. Results show that deploying five SmartNICs with function-based partitioning, which retains packet buffering on the host and offloads header decapsulation, insertion, and lookup-table operations to the SmartNICs, delivers a throughput of 109 MPPS, which is 173% more than a SmartNIC-only setup and reduces latency by 70% compared with a host-only setup. Adaptive power management lowers total power consumption by 12.5% in the optimal partitioning while preserving high throughput and low latency.

As Fifth Generation Networks (5G) networks evolve, operators have an opportunity to create high-quality customer experiences by ensuring seamless and personalized connectivity through consumer-oriented network slicing. However, effectively managing consumer slices remains a challenge due to dynamic user demands, mobility patterns, and the need for real-time Quality of Service (QoS) assurance. This paper presents an Artificial Intelligence (AI)-driven framework for intelligent service feasibility, qualification, and provisioning in 5G networks, to automate consumer slicing. By leveraging real-time data from the Radio Access Network (RAN) and Core Network (CN), along with Service Quality Indicator (SQI), the framework enables dynamic resource allocation, predictive service qualification, and proactive slice optimization. Key innovations of the frame-work include the coherent integration of AI-driven slice load prediction, mobility-aware service provisioning, and automated QoS assurance via Application Programming Interfaces (APIs), ensuring optimal performance for consumer applications while maintaining compliance with Service Level Agreement (SLA). Experimental results demonstrate the framework’s capabilities in enhancing user experiences by enabling personalized slice selection, reducing service disruptions, and optimizing network resource utilization.

We explore a Digital Twin-Based Approach for Smart Manufacturing to improve Sustainability, Efficiency, and Cost-Effectiveness for a steel production plant. Our system is based on a micro-service edge-compute platform that ingests real-time sensor data from the process into a digital twin over a converged network infrastructure. We implement agile machine learning-based control loops in the digital twin to optimize induction furnace heating, enhance operational quality, and reduce process waste. Key to our approach is a Deep Reinforcement learning-based agent used in our machine learning operation (MLOps) driven system to autonomously correlate the system state with its digital twin to identify correction actions that aim to optimize power settings for the plant. We present the theoretical basis, architectural details, and practical implications of our approach to reduce manufacturing waste and increase production quality. We design the system for flexibility so that our scalable event-driven architecture can be adapted to various industrial applications. With this research, we propose a pivotal step towards the transformation of traditional processes into intelligent systems, aligning with sustainability goals and emphasizing the role of MLOps in shaping the future of data-driven manufacturing.

The current P4 programmability model assumes that a P4 programmable device is owned and controlled by a single tenant. However, in typical NFV scenarios, support for multiple tenants is desirable. When each tenant may want to deploy their own P4 pipeline offering different network functions (NF), supporting multiple co-existing tenant pipelines on a single platform is difficult because it requires pipeline merging, control plane support, and resource management of the platform. In this paper, we present P4-MTAGG, a novel framework for flexibly deploying multiple P4 programmable NFs on a programmable match-action pipeline while supporting multiple tenants. P4-MTAGG consists of i) novel compiler-add-ons for automatic merging multiple P4-pipelines, ii) p4runtime-proxy to allow for control plane access of the aggregated pipelines together with policy-based resource management for the P4 target, and iii) orchestrator to automate the provisioning of a network node utilizing aggregation either in a simulated or real hardware environment. In this demo, we show how P4-MTAGG aggregates multiple NFs of varying complexity in Mininet. The user can orchestrate the aggregation process through a GUI. The per-tenant traffic is routed through the set of NFs using segment routing. Through the GUI, the user can instruct the p4runtime-proxy to enforce per-tenant bandwidth limits, which configure the per-tenant available resources in the data plane.

In the realm of cloud native environments, Ku-bernetes has emerged as the de facto orchestration system for containers, and the service mesh architecture, with its interconnected microservices, has become increasingly prominent. Efficient scheduling and resource allocation for these microservices play a pivotal role in achieving high performance and maintaining system reliability. In this paper, we introduce a novel approach for container scheduling within Kubernetes clusters, leveraging Graph Attention Networks (GATs) for representation learning. Our proposed method captures the intricate dependencies among containers and services by constructing a representation graph. The deep Q-learning algorithm is then employed to optimize scheduling decisions, focusing on container-to-node placements, CPU request-response allocation, and adherence to node affinity and anti-affinity rules. Our experiments demonstrate that our GATs-based method outperforms traditional scheduling strategies, leading to enhanced resource utilization, reduced service latency, and improved overall system throughput. The insights gleaned from this study pave the way for a new frontier in cloud native performance optimization and offer tangible benefits to industries adopting microservice-based architectures.

Mobile Edge Clouds (MECs) address the critical needs of bandwidth-intensive, latency-sensitive mobile applications by positioning computing and storage resources at the network's edge in Edge Data Centers (EDCs). However, the diverse, dynamic nature of EDCs' resource capacities and user mobility poses significant challenges for resource allocation and management. Efficient EDC operation requires accurate forecasting of computational load to ensure optimal scaling, service placement, and migration within the MEC infrastructure. This task is complicated by the temporal and spatial fluctuations of computational load.We develop a novel MEC computational demand forecasting method using Federated Learning (FL). Our approach leverages FL's distributed processing to enhance data security and prediction accuracy within MEC infrastructure. By incorporating uncertainty bounds, we improve load scheduling robustness. Evaluations on a Tokyo dataset show significant improvements in forecast accuracy compared to traditional methods, with a 42.04% reduction in Mean Absolute Error (MAE) using LightGBM and a 34.93% improvement with CatBoost, while maintaining minimal networking overhead for model transmission.

Bundling of multiple access technologies is currently being standardized by 3GPP in the 5G access traffic steering, switching and splitting (ATSSS) framework, with the goal to increase robustness, resiliency and capacity of wireless access. A key part of an ATSSS framework is the packet scheduler, which decides the access network over which each packet is to be transmitted. As wireless channels are highly dynamic, a challenge for any scheduler is to correctly estimate the capacity of each path, and thereby avoid congesting the paths. In this paper, we further develop a recent packet scheduler that exploits cross-layer information from the congestion control state of individual transport layer tunnels when making scheduling decisions. Our aim is to achieve good path utilization while keeping the congestion delay low. Extensive emulations show that our approach reduces the excess delay at the bottleneck to as little as 34%. We furthermore show that our approach improves the performance of end-to-end applications including WebRTC and YouTube compared to state-of-the art.

Load forecasting is a crucial topic in energy management systems (EMS) due to its vital role in optimizing energy scheduling and enabling more flexible and intelligent power grid systems. As a result, these systems allow power utility companies to respond promptly to demands in the electricity market. Deep learning (DL) models have been commonly employed in load forecasting problems supported by adaptation mechanisms to cope with the changing pattern of consumption by customers, known as concept drift. A drift magnitude threshold should be defined to design change detection methods to identify drifts. While the drift magnitude in load forecasting problems can vary significantly over time, existing literature often assumes a fixed drift magnitude threshold, which should be dynamically adjusted rather than fixed during system evolution. To address this gap, in this paper, we propose a dynamic drift-adaptive Long Short-Term Memory (DA-LSTM) framework that can improve the performance of load forecasting models without requiring a drift threshold setting. We integrate several strategies into the framework based on active and passive adaptation approaches. To evaluate DA-LSTM in real-life settings, we thoroughly analyze the proposed framework and deploy it in a real-world problem through a cloud-based environment. Efficiency is evaluated in terms of the prediction performance of each approach and computational cost. The experiments show performance improvements on multiple evaluation metrics achieved by our framework compared to baseline methods from the literature. Finally, we present a trade-off analysis between prediction performance and computational costs.

Cloud native computing paradigm allows microservice-based applications to take advantage of cloud infrastructure in a scalable, reusable, and interoperable way. However, in a cloud native system, the vast number of configuration parameters and highly granular resource allocation policies can significantly impact the performance and deployment cost. For understanding and analyzing these implications in an easy, quick, and cost-effective way, we present PerfSim, a discrete-event simulator for approximating and predicting the performance of cloud native service chains in user-defined scenarios. To this end, we proposed a systematic approach for modeling the performance of microservices endpoint functions by collecting and analyzing their performance and network traces. With a combination of the extracted models and user-defined scenarios, PerfSim can then simulate the performance behavior of all services over a given period and provide an approximation for system KPIs, such as requests' average response time. Using the processing power of a single laptop, we evaluated both simulation accuracy and speed of PerfSim in 104 prevalent scenarios and compared the simulation results with the identical deployment in a real Kubernetes cluster. We achieved ~81-99\% simulation accuracy in approximating the average response time of incoming requests and ~16-1200 times speed-up factor for the simulation.

This paper focuses on hybrid pipeline designs for User Plane Function and next-generation NodeB leveraging target-specific features and an insightful discussion of P4 and target challenges and limitations. The entire or disaggregated UPF runs on P4 targets and allocates packet processing data paths in P4 hardware or DPDK/x86 software based on flow characteristics (e.g., heavy hitters) and QoS requirements (e.g., low-latency slices). For the hybrid gNodeB, most packet processing is executed in commodity Tofino hardware, while unsupported functions such as Automatic Repeat Request and cryptography are performed in DPDK/x86. We show that our hybrid UPF improves the scalability by 18× and reduces latency up to 50%. The results also suggest that careful traffic allocation to pipeline targets is required to optimize each target's strength and avoid processing delays. Finally, we demonstrate a QoS-oriented application of the hybrid UPF and present gNodeB buffer service benchmarks.

The optimal placement of virtual network functions (VNFs) improves the overall performance of service function chains (SFCs) and decreases the operational costs for mobile network operators. To cope with changes in demands, VNF instances may be added or removed dynamically, resource allocations may be adjusted, and servers may be consolidated. To maintain an optimal placement of SFCs when conditions change, SFC reconfiguration is required, including the migration of VNFs and the rerouting of service-flows. However, such reconfigurations may lead to stress on the VNF infrastructure, which may cause service degradation. On the other hand, not changing the placement may lead to suboptimal operation, and servers and links may become congested or underutilized, leading to high operational costs. In this paper, we investigate the trade-off between the reconfiguration of SFCs and the optimality of their new placement and service-flow routing. We develop a multi-objective genetic algorithm that explores the Pareto front by balancing the optimality of the new placement and the cost to achieve it. Our numerical evaluations show that a small number of reconfigurations can significantly reduce the operational cost of the VNF infrastructure. In contrast, too much reconfiguration may not pay off due to high costs. We believe that our work provides an important tool that helps network providers to plan a good reconfiguration strategy for their service chains.

Networked systems have recently aimed to use multiple access networks in parallel to increase resiliency, availability and capacity. However, different paths may have different latency characteristics, which may lead to out-of-order packet delivery. This may severely impact both the end-to-end application performance and the capacity utilisation of multiaccess systems. In this paper, we show that in-network support for packet reordering for multiaccess systems that are based on multiple transport layer tunnels is beneficial for several application types. Our findings are applicable to TCP and QUIC traffic in the 3GPP ATSSS context, where we use the MP-DCCP tunneling framework with a buffer-based packet reordering approach that uses a dynamic timing threshold to cope with variation of path delays over time. We demonstrate achievable performance gains for a wide range of path latency differences and end-to-end round trip times when using different in-network reordering algorithms.

The emergence of Network Functions Virtualization (NFV) is being heralded as an enabler of the recent technologies such as 5G/6G, IoT and heterogeneous networks. Existing NFV monitoring frameworks either do not have the capabilities to express the range of telemetry items needed to perform management or do not scale to large traffic volumes and rates. We present IntOpt, a scalable and expressive telemetry system designed for flexible NFV monitoring using active probing and P4. IntOpt allows us to specify monitoring requirements for individual service chain, which are mapped to telemetry item collection jobs that fetch the required telemetry items from P4 programmable data-plane elements. We propose mixed integer linear program (MILP) as well as a simulated annealing based random greedy (SARG) meta-heuristic approach to minimize the overhead due to active probing and collection of telemetry items. Using P4-FPGA, we benchmark the overhead for telemetry collection. Our numerical evaluation shows that the proposed approach can reduce monitoring overheads by 39% and monitoring delays by 57%. Such optimization may as well enable existing expressive monitoring frameworks to scale for larger real-time networks.

Next-generation cellular networks, i.e., beyond fifth generation (B5G) and sixth generation (6G) networks, aim at significantly increasing capacity by deploying a massive number of small cells (SCs) and leveraging millimeter wave (mmWave) links that have large bandwidth availability. However, as optical connections to a large number of base stations (BSs) are not cost effective, wireless backhaul (BH) links may be used to connect SCs to the core network using mmWave frequencies and forming multihop mesh BH paths. While very flexible in deployment, the uncertainty in user demand, which varies over time and place, makes network planning and management challenging. In this work, we propose novel algorithms to solve the joint problem of energy-efficient user association, BH traffic routing and BS/BH link on/off switching. We first formulate an exact model assuming user traffic demand uncertainty leveraging Γ-robustness theory. As the model is intractable for large-scale instances, we develop a novel greedy robust heuristic (P-HEUR), which includes a robustifying step to effectively cope with demand uncertainty. Our evaluation demonstrates that P-HEUR provides robust and energy-efficient solutions quickly for different scenarios and traffic demand uncertainty levels, and it significantly outperforms state-of-the-art approaches while achieving up to 83% of the optimal capacity, even in the most demanding scenarios in terms of traffic load and demand uncertainty, with up to 200k times lower complexity.

The dynamicity of real-world systems poses a significant challenge to deployed predictive machine learning (ML) models. Changes in the system on which the ML model has been trained may lead to performance degradation during the system’s life cycle. Recent advances that study non-stationary environments have mainly focused on identifying and addressing such changes caused by a phenomenon called concept drift. Different terms have been used in the literature to refer to the same type of concept drift and the same term for various types. This lack of unified terminology is set out to create confusion on distinguishing between different concept drift variants. In this paper, we start by grouping concept drift types by their mathematical definitions and survey the different terms used in the literature to build a consolidated taxonomy of the field. We also review and classify performance-based concept drift detection methods proposed in the last decade. These methods utilize the predictive model’s performance degradation to signal substantial changes in the systems. The classification is outlined in a hierarchical diagram to provide an orderly navigation between the methods. We present a comprehensive analysis of the main attributes and strategies for tracking and evaluating the model’s performance in the predictive system. The paper concludes by discussing open research challenges and possible research directions.

Cloud computing offers a wide range of services through a pool of heterogeneous Physical Machines (PMs) hosted on cloud data centers, where each PM can host several Virtual Machines (VMs). Resource sharing among VMs comes with major benefits, but it can create technical challenges that have a detrimental effect on the performance. To ensure a specific service level requested by the cloud-based applications, there is a need for an approach to assign adequate resources to each VM. To this end, we present our novel Multi-Loop Control approach, calledMultiScaler, to allocate resources to VMs based on the Service Level Agreement (SLA) requirements and the run-time conditions.MultiScaleris mainly composed of three different levels working closely with each other to achieve an optimal resource allocation. We propose a set of tailor-made controllers to monitor VMs and take actions accordingly to regulate contention among collocated VMs, to reallocate resources if required, and to migrate VMs from one PM to another. The evaluation in a VMware cluster have shown that theMultiScalerapproach can meet applications performance goals and guarantee the SLA by assigning the exact resources that the applications require. Compared with sophisticated baselines,MultiScalerproduces significantly better reaction to changes in workloads even under the presence of noisy neighbors.

Emerging distributed applications, such as big data analytics, generate a large number of flows that concurrently transport data across data center networks. To improve their performance, it is required to account for the behavior of such a collection of flows, i.e., coflows, rather than individual ones. State-of-the-art solutions achieve near-optimal completion time by continuously reordering unfinished coflows at the end-host and using network priorities.This paper shows that dynamically changing flow priorities at the end-host, without considering in-flight packets, can cause high degrees of packet reordering, thus imposing pressure on the congestion control and potentially harming network performance in the presence of switches with shallow buffers. We present pCoflow, a new solution that integrates end-host based coflow ordering with in-network scheduling based on packet history. Our evaluation shows that pCoflow improves in coflow completion time upon state-of-the-art solutions by up to 34% for varying loads.

Industrial Internet of Things (I-IoT) applications require a large number of sensor data to be processed under tight delay and jitter constraints. In such applications, flexible event detection and fast reaction to critical events is an important building block. Traditional approaches use either proprietary networks and dedicated hardware or transmit sensor data towards processing elements in the Cloud or at the Network Edge, using distributed stream processing frameworks. For scalability, a large number of servers are needed and processing on commodity CPUs typically involves high and unpredictable latency. In this article, we explore how programmable data planes can be used to detect events flexibly and trigger customized and programmable actions directly from the switch program or the programmable network interface card (SmartNIC). We present FastReact-PS, an event-based publish/subscribe I-IoT processing framework in P4 language, which can be flexibly customized from the control plane. Together with stateful processing, FastReact-PS supports windowed time series analysis as well as complex event detection and processing based on Boolean logic directly in the data plane of newly emerging programmable networking devices. The logic can be adjusted dynamically from the control plane without the need for recompilation. We implement FastReact-PS in P4 and evaluate it on both a SmartNIC and a DPDK-based software switch running in user space. Our evaluation shows that the latency is reduced by one order of magnitude compared to end-host based approaches at significantly lower jitter while being scalable to processing up to 11 million events per second.

Network Function Virtualization (NFV) becomes the primary driver for the evolution of 5G networks, and in recent years, Network Function Cloudification (NFC) proved to be an inevitable part of this evolution. Microservice architecture also becomes the de facto choice for designing a modern Cloud Native Network Function (CNF) due to its ability to decouple components of each CNF into multiple independently manageable microservices. Even though taking advantage of microservice architecture in designing CNFs solves specific problems, this additional granularity makes estimating resource requirements for a Production Environment (PE) a complex task and sometimes leads to an over-provisioned PE. Traditionally, performance engineers dimension each CNF within a Service Function Chain (SFC) in a smaller Performance Testing Environment (PTE) through a series of performance benchmarks. Then, considering the Quality of Service (QoS) constraints of a Service Provider (SP) that are guaranteed in the Service Level Agreement (SLA), they estimate the required resources to set up the PE. In this paper, we used a machine learning approach to model the impact of each microservice's resource configuration (i.e., CPU and memory) on the QoS metrics (i.e. serving throughput and latency) of each SFC in a PTE. Then, considering an SP's Service Level Objectives (SLO), we proposed an algorithm to predict each microservice's resource capacities in a PE. We evaluated the accuracy of our prediction on a prototype of a cloud native 5G Home Subscriber Server (HSS). Our model showed 95%-78% accuracy in a PE that has 2-5 times more computing resources than the PTE.

Abstract Network Function Virtualization (NFV) is an emerging technology to consolidate network functions onto high volume storages, servers and switches located anywhere in the network. Virtual Network Functions (VNFs) are chained together to provide a specific network service, called Service Function Chains (SFCs). Regarding to Quality of Service (QoS) requirements and network features and states, SFCs are served through performing two tasks: VNF placement and link embedding on the substrate networks. Reducing deployment cost is a desired objective for all service providers in cloud/edge environments to increase their profit form demanded services. However, increasing resource utilization in order to decrease deployment cost may lead to increase the service latency and consequently increase SLA violation and decrease user satisfaction. To this end, we formulate a multi-objective optimization model to joint VNF placement and link embedding in order to reduce deployment cost and service latency with respect to a variety of constraints. We, then solve the optimization problem using two heuristic-based algorithms that perform close to optimum for large scale cloud/edge environments. Since the optimization model involves conflicting objectives, we also investigate pareto optimal solution so that it optimizes multiple objectives as much as possible. The efficiency of proposed algorithms is evaluated using both simulation and emulation. The evaluation results show that the proposed optimization approach succeed in minimizing both cost and latency while the results are as accurate as optimal solution obtained by Gurobi (5%).

Future mobile traffic growth will require 5G cellular networks to densify the deployment of small cell base stations (BS). As it is not feasible to form a backhaul (BH) by wiring all BSs to the core network, directional mmWave links can be an attractive solution to form BH links, due to their large available capacity. When small cells are powered on/off or traffic demands change, the BH may require reconfiguration, leading to topology and traffic routing changes. Ideally, such reconfiguration should be seamless and should not impact existing traffic. However, when using highly directional BH antennas which can be dynamically rotated to form new links, this can become time- consuming, requiring the coordination of BH interface movements, link establishment and traffic routing. In this paper, we propose greedy-based heuristic algorithms to solve the BH reconfiguration problem in real-time. We numerically compare the proposed algorithms with the optimal solution obtained by solving a mixed integer linear program (MILP) for smaller instances, and with a sub- optimal reduced MILP for larger instances. The obtained results indicate that the greedy-based algorithms achieve good quality solutions with significantly decreased execution time.

Managing and scaling virtual network function (VNF) service chains require the collection and analysis of network statistics and states in real time. Existing network function virtualization (NFV) monitoring frameworks either do not have the capabilities to express the range of telemetry items needed to perform management or do not scale to large traffic volumes and rates. We present IntOpt, a scalable and expressive telemetry system designed for flexible VNF service chain network monitoring using active probing. IntOpt allows to specify monitoring requirements for individual service chain, which are mapped to telemetry item collection jobs that fetch the required telemetry items from P4 (programming protocol-independent packet processors) programmable dataplane elements. In our approach, the SDN controller creates the minimal number of monitoring flows to monitor the deployed service chains as per their telemetry demands in the network. We propose a simulated annealing based random greedy metaheuristic (SARG) to minimize the overhead due to active probing and collection of telemetry items. Using P4-FPGA, we benchmark the overhead for telemetry collection and compare our simulated annealing based approach with a naïve approach while optimally deploying telemetry collection probes. Our numerical evaluation shows that the proposed approach can reduce the monitoring overhead by 39% and the total delays by 57%. Such optimization may as well enable existing expressive monitoring frameworks to scale for larger real-time networks.

Cloud computing is gaining significant traction and virtualized data centers are becoming popular as a cost-effective infrastructure in telecommunication industry. Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS) are being widely deployed and utilized by end users, including many private as well as public organizations. Despite its wide-spread acceptance, security is still the biggest threat in cloud computing environments. Users of cloud services are under constant fear of data loss, security breaches, information theft and availability issues. Recently, learning-based methods for security applications are gaining popularity in the literature with the advents in machine learning (ML) techniques. In this work, we explore applicability of two well-known machine learning approaches, which are, Artificial Neural Networks (ANN) and Support Vector Machines (SVM), to detect intrusions or anomalous behavior in the cloud environment. We have developed ML models using ANN and SVM techniques and have compared their performances. We have used UNSW-NB-15 dataset to train and test the models. In addition, we have performed feature engineering and parameter tuning to find out optimal set of features with maximum accuracy to reduce the training time and complexity of the ML models. We observe that with proper features set, SVM and ANN techniques have been able to achieve anomaly detection accuracy of 91% and 92% respectively, which is higher compared against that of the one achieved in the literature, with reduced number of features needed to train the models.

Network Function Virtualization (NFV) with minimum-delay service function chains will be a key enabling technology for next generation mobile networks, such as 5G. In this paper, we present a new approach to generate problem instances for the cost optimized delay sensitive Virtualized Network Function (VNF) placement and routing problem, where we know by construction the optimal solution for the given objective function. Our approach produces problem instances, which can be used to test and benchmark heuristic algorithms against. We then implement an Affinity based simulated annealing (ABSA) heuristic approach for cost optimized delay aware placement of VNFs along with a set of greedy approaches. We evaluate the implemented approaches against the optimal solution by generating several problem instances using our proposed method.

Mobile nodes are typically equipped with multiple radios and can connect to multiple radio access networks (e.g. WiFi, LTE and 5G). Consequently, it is important to design mechanisms that efficiently manage multiple network interfaces for aggregating the capacity, steering of traffic flows or switching flows among multiple interfaces. While such multi-access solutions have the potential to increase the overall traffic throughput and communication reliability, the variable latencies on different access links introduce packet delay variation which has negative effect on the application quality of service and user quality of experience. In this paper, we present a new IP-compatible multipath framework for heterogeneous access networks. The framework uses Multipath Datagram Congestion Control Protocol (MP-DCCP) - a set of extensions to regular DCCP - to enable a transport connection to operate across multiple access networks, simultaneously. We present the design of the new protocol framework and show simulation and experimental testbed results that (1) demonstrate the operation of the new framework, and (2) demonstrate the ability of our solution to manage significant packet delay variation caused by the asymmetry of network paths, by applying pluggable packet scheduling or reordering algorithms.

Network Function Virtualization (NFV) is an emerging technology to consolidate network functions onto high volume storages, servers and switches located anywhere in the network. Virtual Network Functions (VNFs) are chained together to provide a specific network service. Therefore, an effective service chain placement strategy is required to optimize the resource allocation and consequently to reduce the operating cost of the substrate network. To this end, we propose four genetic-based algorithms using roulette wheel and tournament selection techniques in order to place service chains considering two different placement strategies. Since mapping of service chains sequentially (One-at-a-time strategy) may lead to suboptimal placement, we also propose Simultaneous strategy that places all service chains at the same time to improve performance. Our goal in this work is to reduce deployment cost of VNFs while satisfying constraints. We consider Geant network as the substrate network along with its characteristics extracted from SndLib. The proposed algorithms are able to place service chains with any type of service graph. The performance benefits of the proposed algorithms are highlighted through extensive simulations.

Future mobile data traffic predictions expect a significant increase in user data traffic, requiring new forms of mobile network infrastructures. Fifth generation (5G) communication standards propose the densification of small cell access base stations (BSs) in order to provide multigigabit and low latency connectivity. This densification requires a high capacity backhaul network. Using optical links to connect all the small cells is economically not feasible for large scale radio access networks where multiple BSs are deployed. A wireless backhaul formed by a mesh of millimeter-wave (mmWave) links is an attractive mobile backhaul solution, as flexible wireless (multihop) paths can be formed to interconnect all the access BSs. Moreover, a wireless backhaul allows the dynamic reconfiguration of the backhaul topology to match varying traffic demands or adaptively power on/off small cells for green backhaul operation. However, conducting and precisely controlling reconfiguration experiments over real mmWave multihop networks is a challenging task. In this paper, we develop a Software-Defined Networking (SDN) based approach to enable such a dynamic backhaul reconfiguration and use real-world mmWave equipment to setup a SDN-enabled mmWave testbed to conduct various reconfiguration experiments. In our approach, the SDN control plane is not only responsible for configuring the forwarding plane but also for the link configuration, antenna alignment, and adaptive mesh node power on/off operations. We implement the SDN-based reconfiguration operations in a testbed with four nodes, each equipped with multiple mmWave interfaces that can be mechanically steered to connect to different neighbors. We evaluate the impact of various reconfiguration operations on existing user traffic using a set of extensive testbed measurements. Moreover, we measure the impact of the channel assignment on existing traffic, showing that a setup with an optimal channel assignment between the mesh links can result in a 44% throughput increase, when compared to a suboptimal configuration.

Providing network reliability as well as low and predictable latency is important especially for Industrial Automation and Control Networks. However, diagnosing link status from the control plane has high latency and overhead. In addition, the communication with the industrial controller may impose additional network latency. We present FastReact - a system enabling In-Network monitoring, control and caching for Industrial Automation and Control Networks. FastReact outsources simple monitoring and control actions to evolving programmable data planes using the P4 language. As instructed by the Industrial Controller through a Northbound API, the SDN controller composes control actions using Boolean Logic which are then installed in the data plane. The data plane parses and caches sensor values and performs simple calculations on them which are connected to fast control actions that are executed locally. For resiliency, FastReact monitors liveness and response of sensors/actuators and performs a fast local link repair in the data plane if a link failure is detected. Our testbed measurement show that FastReact can reduce the sensor/actuator delay while being resilient against several failure events.

Unmanned aerial vehicles (UAVs) have gained a lot of popularity in diverse wireless communication fields. They can act as high- altitude flying relays to support communications between ground nodes due to their ability to provide line-of- sight links. With the flourishing Internet of Things, several types of new applications are emerging. In this paper, we focus on bandwidth hungry and delay-tolerant applications where multiple pairs of transceivers require the support of UAVs to complete their transmissions. To do so, the UAVs have the possibility to employ two different bands namely the typical microwave and the high-rate millimeter wave bands. In this paper, we develop a generic framework to assign UAVs to supported transceivers and optimize their trajectories such that a weighted function of the total service time is minimized. Taking into account both the communication time needed to relay the message and the flying time of the UAVs, a mixed non-linear programming problem aiming at finding the stops at which the UAVs hover to forward the data to the receivers is formulated. An iterative approach is then developed to solve the problem. First, a mixed linear programming problem is optimally solved to determine the path of each available UAV. Then, a hierarchical iterative search is executed to enhance the UAV stops' locations and reduce the service time. The behavior of the UAVs and the benefits of the proposed framework are showcased for selected scenarios.

With the emergence of millimeter-Wave (mmWave) communication technology, the capacity of mobile backhaul networks can be significantly increased. On the other hand, Mobile Edge Computing (MEC) provides an appropriate infrastructure to offload latency-sensitive tasks. However, the amount of resources in MEC servers is typically limited. Therefore, it is important to intelligently manage the MEC task offloading by optimizing the backhaul bandwidth and edge server resource allocation in order to decrease the overall latency of the offloaded tasks. This paper investigates the task allocation problem in MEC environment, where the mmWave technology is used in the backhaul network. We formulate a Mixed Integer NonLinear Programming (MINLP) problem with the goal to minimize the total task serving time. Its objective is to determine an optimized network topology, identify which server is used to process a given offloaded task, find the path of each user task, and determine the allocated bandwidth to each task on mmWave backhaul links. Because the problem is difficult to solve, we develop a two-step approach. First, a Mixed Integer Linear Program (MILP) determining the network topology and the routing paths is optimally solved. Then, the fractions of bandwidth allocated to each user task are optimized by solving a quasi-convex problem. Numerical results illustrate the obtained topology and routing paths for selected scenarios and show that optimizing the bandwidth allocation significantly improves the total serving time, particularly for bandwidth-intensive tasks.

Future 5G mobile networks will require increased backhaul (BH) capacity to connect a massive amount of high capacity small cells (SCs) to the network. Because having an optical connection to each SC might be infeasible, mmWave-based (e.g. 60 GHz) BH links are an interesting alternative due to their large available bandwidth. To cope with the increased path loss, mmWave links require directional antennas that should be able to direct their beams to different neighbors, to dynamically change the BH topology, in case new nodes are powered on/off or the traffic demand has changed. Such BH adaptation needs to be orchestrated to minimize the impact on existing traffic. This paper develops a Software-defined networking-based framework that guides the optimal reconfiguration of mesh BH networks composed by mmWave links, where antennas need to be mechanically aligned. By modelling the problem as a Mixed Integer Linear Program (MILP), its solution returns the optimal ordering of events necessary to transition between two BH network configurations. The model creates backup paths whenever it is possible, while minimizing the packet loss of ongoing flows. A numerical evaluation with different topologies and traffic demands shows that increasing the number of BH interfaces per SC from 2 to 4 can decrease the total loss by more than 50%. Moreover, when increasing the total reconfiguration time, additional backup paths can be created, consequently reducing the reconfiguration impact on existing traffic.

Geo-distributed Data Centers (DCs) are increasingly common in order to provide scalability for increasing compute demands of modern applications. When multiple geo-distributed DCs serve user requests, it is important to determine which DC and server to select to fulfill the demand at minimum cost, given that enough resources are available in terms of e.g., CPU and bandwidth. This is a complex task since every DC has different operational costs due to e.g. energy, carbon emission, and bandwidth costs. In this paper, we develop a novel mathematical optimization model that guides the decision maker which DC to select, which server to use, and which DC gateway and network path to use to route the user demand in order to satisfy the time varying compute, bandwidth, and latency demands. Our model is based on the concept of virtual networks, which have different requirements in terms of e.g. latency, and we model the queuing delay as a function of the traffic load. Our extensive numerical evaluation, which is based on real-world DC locations, their resource provision costs, and typical demand patterns, shows how operational costs increase with the traffic load, and we analyze the impact of different latency bounds on the performance of different virtual networks.

In virtualized datacenters (vDCs), dynamic consolidation of virtual machines (VMs) is used as one of the most common techniques to achieve both energy-and resource-utilization efficiency. Live migrations of VMs are used for dynamic consolidation but due to dynamic resource demand variation of VMs may lead to frequent and non-optimal migrations. Assuming deterministic workload of the VMs may ensure the most energy/resource-efficient VM allocations but eventually may lead to significant resource contention or under-utilization if the workload varies significantly over time. On the other hand, adopting a conservative approach by allocating VMs depending on their peak demand may lead to low utilization, if the peak occurs infrequently or for a short period of time. Therefore, in this work we design a robust VM migration scheme that strikes a balance between protection for resource contention and additional energy costs due to powering on more servers while considering uncertainties on VMs resource demands. We use the theory of Gamma robustness and derive a robust Mixed Integer Linear programming (MILP) formulation. Due to the complexity, the problem is hard to solve for online optimization and we propose a novel heuristic based on Tabu Search. Using several scenarios, we show that that the proposed heuristic can achieve near optimal solution qualities in a short time and scales well with the instance sizes. Moreover, we quantitatively analyze the trade-off between energy cost versus protection level and robustness.

OpenStack has been widely acknowledged to be one of the most important open source cloud platforms. In order to perform experimentally driven research in the area of cloud and cloud networking, there is however a big gap, because most researchers do not have access to a large cloud deployment and cannot change networking or compute infrastructure in order to test their algorithms and protocols on a large-scale. We developed OpenStackEmu, which is to the best of our knowledge the first attempt that combines OpenStack infrastructure with a Software Defined Networking (SDN) based controller such as OpenDay-light and a large-scale network emulator CORE (Common Open Research Emulator). The OpenStack compute and control nodes are connected to the CORE emulation server using TUN/TAP interfaces that inject the control (e.g. for VM migration) and data (VM-to-VM traffic) packets into a customizable network topology that is emulated using configurable Open vSwitches using CORE emulator. Experimenters can define e.g. fat-tree or distributed data center topologies and study the behavior of real VMs and services in those VMs under different background loads and SDN routing policies. We integrated the data center traffic generator DCT2Gen that allows to generate realistic background traffic based on traces from real data centers. Experimenters can study the performance impact of different VM migration strategies or different routing and load balancing schemes on real VM and application performance using different emulated topologies. We believe that OpenStackEmu is an important tool for both the SDN and OpenStack community in order to evaluate the performance of novel algorithms and protocols in the area of cloud networking.

With the advance of fifth generation (5G) networks, network density needs to grow significantly in order to meet the required capacity demands. A massive deployment of small cells may lead to a high cost for providing fiber connectivity to each node. Consequently, many small cells are expected to be connected through wireless links to the umbrella eNodeB, leading to a mesh backhaul topology. This backhaul solution will most probably be composed of high capacity point-to-point links, typically operating in the millimeter wave (mmWave) frequency band due to its massive bandwidth availability. In this paper, we propose a mathematical model that jointly solves the user association and backhaul routing problem in the aforementioned context, aiming at the energy efficiency maximization of the network. Our study considers the energy consumption of both the access and backhaul links, while taking into account the capacity constraints of all the nodes as well as the fulfillment of the service-level agreements (SLAs). Due to the high complexity of the optimal solution, we also propose an energy efficient heuristic algorithm (Joint), which solves the discussed joint problem, while inducing low complexity in the system. We numerically evaluate the algorithm performance by comparing it not only with the optimal solution but also with reference approaches under different traffic load scenarios and backhaul parameters. Our results demonstrate that Joint outperforms the state-of-the-art, while being able to find good solutions, close to optimal, in short time.

In order to increase the capacity of next generation mobile networks, network densification is a key aspect to provide better coverage and increased data rates to end users. Network operators are thinking to deploy wireless backhaul solutions to cut down cabling costs for connecting the small cell nodes. Consequently, the next generation mobile network architecture may contain a massive amount of small cells that are connected through wireless backhaul links, forming mesh or tree structures towards the core network, under the umbrella coverage of eNodeB type macro cells. In this paper, we use a Software-Defined Networking (SDN) based architecture for the operation and management of such small cell backhaul networks. By extending OpenFlow, the SDN controller is able to reconfigure not only the routing, but also the wireless backhaul configuration, such as channel assignment to backhaul links. We demonstrate the effectiveness of our approach by using testbed measurements and show that, when using our SDN-based reconfiguration, the network downtime for existing traffic due to channel re-assignment is significantly reduced, when comparing to a distributed routing based approach. Moreover, by using SDN based fast-failover techniques, we can instantaneously redirect traffic to other neighbors when links fail or neighbors are powered down, without the need for controller interaction.

A dense deployment of small cells is one of the key characteristics envisioned for future 5G mobile networks in order to provide the required capacity increase. To cut cabling costs, small cells are foreseen to form multihop topologies using high capacity backhaul wireless links in the mmWave bands. Such small cells are deployed within the coverage area of a macro cell (eNB) to provide localized capacity where needed according to the well known Heterogeneous Network concept (HetNet). However, green networking will be very important because powering on unnecessarily a massive amount of small cells may lead to increased OPEX and CO2 emission. In this paper, we propose an optimization model that minimizes the total power consumption of 5G HetNets deployments while providing the required capacity and coverage. Our model jointly optimizes the user association, routing in the multihop backhaul and decides to power on or off the small cells to serve the user demands. Our numerical evaluation show significant power savings over a large range of traffic demand distributions while keeping the blocking probability low.

Recently, 5G systems started to explore the usage of mmWave bands both for access and backhaul. Cellular systems operating at such frequencies would achieve a significantly higher capacity than current systems due to the large frequency block available. Due to the propagation characteristics, highly directional mmWave links may frequently switch between Line-of-sight, Non Line-of-sight and outage if the beam is blocked by obstacles. Such intermittent connectivity combined with high throughput causes TCP to perform very badly. In this paper, we study the performance of several TCP variants over intermittent mmWave links. We focus on latency aspects, use ns-3 with extensions for mmWave links and evaluate the TCP performance with real Linux TCP stacks using the Direct Code Execution (DCE) framework. We also investigate the benefit of bufferbloat solutions such as CoDel, with different TCP variants. We investigate the impact of CoDels configuration parameters and different RTTs on TCP throughput, fairness and queueing delay. The problem of slow recovery of TCP after NLOS periods is less pronounced when multiple flows coexist under CoDel only if some of the flows have very short RTTs. However, under presence of NLOS, the fairness between different flows with different RTT values is reduced.

Fifth generation (5G) wireless networks will target at energy and spectrum efficient solutions to cope with the increasing demands in capacity and energy efficiency. To achieve this joint goal, dense networks of small cells (SCs) are expected to overlay the existing macro cells. In parallel, for the SC connection to the core network, a promising solution lies in a mesh network of high capacity millimeter wave backhaul (BH) links. In the considered 5G architecture, each SC is able to forward its BH traffic to the core network through alternative paths, thus offering high BH network reliability. In this context, the joint problem of user association and BH routing becomes challenging. In this paper, we focus on this problem targeting at energy and spectrum efficient solutions. A low-complexity algorithm is proposed, which bases its user association and BH routing decision i) on minimizing the spectrum resources to guarantee the user rate, so as to provide high spectrum efficiency, and ii) on minimizing both the access network and BH route power consumption to provide high energy efficiency. Our results show that our solution provides better trade-offs between energy and spectrum efficiency than the state-of-the-art in 3GPP scenarios.

In virtualized datacenters (vDCs), dynamic consolidation of virtual machines (VMs) is used to achieve both energy-efficiency and load balancing among different physical machines (PMs). Using VM live migrations, we can consolidate VMs on a smaller number of hosts to power down unused PMs and save energy. Most migration schemes are however oblivious to the characteristics of services that run inside VMs, and thus may lead to migrations where VMs competing for the same resource type are packed on the same PM. As a result, VMs may suffer from significant resource contention and noticeable degradation in their performance. Using resource sensitivity values of VMs (i.e., quantitative measures to reflect how much a VM is sensitive to its requested resources such as CPU, Mem, and Disk), we have designed a novel VM consolidation approach to optimize placement of VMs on available PMs. We validated our approach using five well-known applications/benchmarks with various resource demand signatures: varying from pure CPU/Mem/Disk-intensive to mixtures of them. Our extensive numerical evaluation illustrates that, for the same power consumption, our approach improve the performance of cloud services by 9 - 12%, on average, when compared with current sensitivity oblivious approaches.

Spectrum scarcity together with high capacity demands make the use of millimeter wave (mmWave) frequencies an interesting alternative for next generation, i.e., fifth generation (5G), networks. Although mmWave is expected to play a key role for both access network and backhaul (BH), its initial use in the BH network seems more straight-forward. This stems from the fact that, in the BH case, its deployment is less challenging due to the fixed locations of BH transceivers. Still, provided that mmWave spectrum consists of several subbands, each one with different characteristics and thus different deployment constraints (e.g., channel bandwidth, maximum transmission power), a comparison is required in order to gain a better insight into the potentials of each solution. To that end, in this paper, the main mmWave candidate frequency bands are compared in terms of range, throughput and energy consumption. In our results, the bandwidth availability, the maximum transmission power as well as the antenna gains of each BH technology are taken into account, as defined by the Federal Communications Commission. The results are also compared with current industry-oriented state-of-the-art transceiver characteristics in order to gain further insights into the maximum achievable gains of each subband.

Future Access and Backhaul Networks need to significantly carry more data traffic, which calls for network densification. However, when many small cells are densely deployed, the backhaul network might become the bottleneck, as it is economically infeasible to wire every small cell or power them on constantly. Therefore, an effective configuration of a Small Cell Wireless Backhaul Network is important due to its multi-hop operation, possibly operating in mmWave bands. In this paper, we propose to use Software-Defined Networking for the Operation and Management of Small Cell Wireless Backhaul Networks. A SDN controller based in OpenDaylight is in charge of adaptively powering on/off Small Cells and reconfiguring the backhaul forwarding topology, according to traffic demands. The SDN controller uses an optimizer module which can be configured with different policies in order to minimize the energy cost or optimize capacity or latency. We extend OpenFlow in order to gather wireless and power consumption statistics, which are exchanged between Controller and Small Cell Backhaul nodes using a LTE Out-Of-Band Control Channel.

Due to the tremendous increase in traffic demand, current mobile access and backhaul networks face a capacity problem. To increase the capacity, one can deploy many small cells, which may be dynamically powered on where traffic demand arises. As a consequence, the backhaul network for dense small cell deployments needs to cope with the massive increase in user demands but rolling out fiber to each small cell is economically not viable. Recently, mmWave based mobile backhaul networks have raised a lot of interest because of the large chunk of available spectrum in the 60 GHz, and 70/80 GHz band. However, due to the specific propagation characteristic of mmWave spectrum, such links may face frequent outage. In addition, a more flexible design of the backhaul network is desired in order to cope with the diversification of service requirements. In this paper, we propose a small cell backhaul network architecture which is based on the concept of Software Defined Networking. In order to cope with the dynamics of mmWave links, we propose that the SDN control plane calculates for each backhaul link a set of backup links. Using OpenFlow Fast Failover groups, a fast local repair of a mmWave backhaul link can be achieved leading to a resilient backhaul mesh architecture. An evaluation of our concept in a network emulator demonstrates the effectiveness of our approach. Particularly how it leads to lower packet loss and consequently higher throughput.

Live Virtual Machine (VM) migrations are an important tool that is used in modern data centers in order to e.g. Consolidate server racks for maintenance or optimize VM placements across physical hosts. However, live VM migration causes a lot of network stress due to the potential large volume of data that is transmitted between the physical hosts, which may negatively impact other latency sensitive VM to VM traffic. As VM downtime and the time to migrate depend on the allocated resources for migration traffic, it is important to manage the network resources for live VM migration traffic. In this work, we improve the performance for both live VM migration traffic and VM to VM communication using three strategies. First, we take advantage out of the path diversity available in modern data centers and utilize multipath TCP (MPTCP) for live VM traffic. Second, we implement flexible use of queue management strategies such as FQ CODEL or Hierarchy Token Bucket (HTB). Finally, we orchestrate the process into Open Stack Neutron and connect it together with an SDN control application, which runs on Open Daylight. An extensive evaluation in our Open Stack test bed using different VM workload patterns and VM sizes shows, that FQ CODEL can bring down VM to VM latency during ongoing migrations while MPTCP effectively aggregates bandwidth of multiple paths to reduce live VM migration latency and downtime.

Reducing the energy consumption of data centers and the Cloud is very important in order to lower CO_2 footprint and operational cost (OPEX) of a Cloud operator. To this extent, it becomes crucial to minimise the energy consumption by consolidating the number of powered-on physical servers that host the given virtual machines (VMs). In this work, we propose a novel approach to the energy efficient VM consolidation problem by applying Robust Optimisation Theory. We develop a mathematical model as a robust Mixed Integer Linear Program under the assumption that the input to the problem (e.g. resource demands of the VMs) is not known precisely, but varies within given bounds. A numerical evaluation shows that our model allows the Cloud Operator to tradeoff between the power consumption and the protection from more severe and unlikely deviations of the uncertain input.

The Configuration and Management of large WLAN deployments is a challenge and available tools to ease such deployments and introduce new services are either commercial or very inflexible. In this paper, we present a different approach to such challenges called QoS enabled CloudMAC, which is to the best of our knowledge the first step towards QoS enabled WiFi MAC layer processing as an example of Network Function Virtualization. By moving the MAC layer processing to the cloud and integrating our architecture with QoS aware OpenFlow deployment, a software defined networking approach, we achieve a new level of flexibility, control and reconfigurability. CloudMAC Access Points (AP) just forward MAC layer frames towards a set of VMs (Virtual Access Points - VAP) that are responsible for processing MAC layer data and management frames (such as beacons, probe requests, etc). We have extended the SDN that connects the VAPs with the physical APs to support different packet prioritisation strategies such as HTB, SFQ, or FQ_CoDel. Our SDN controller is based on OpenDaylight which creates layer 2 forwarding rules that effectively prioritise CloudMAC traffic over legacy traffic. Our evaluation in a real testbed shows that packet prioritization strategies, especially FQ_CoDel, lead to good throughput and low latency for CloudMAC traffic while at the same time maintaining low latency for background traffic.

Abstract In Wireless mesh networks mesh access points (MAPs) forward traffic wirelessly towards users or Internet gateways. A user device usually connects to the MAP with the strongest signal, as such MAP should guarantee the best quality of service. However, this connection policy may lead to: (i) unfairness towards users that are distant from gateways; (ii) uneven distribution of users to MAPs; and (iii) inefficient use of network paths. We present a new model and solution approach to the problem of assigning users to MAPs and routing the data within the mesh network with the objective of providing max–min fair throughput. The problem is formulated as a mixed‐integer linear programming problem (MILP). Because of the inherent complexity of the problem, real size instances cannot be solved to optimality within the time limits for online optimization. Therefore, we propose an original heuristic solution algorithm for the resulting MILP. Both numerical comparisons and network simulations demonstrate the effectiveness of the proposed heuristic. For random networks, the heuristic achieves 98% of the optimal solution. Network simulations show that in medium‐sized networks, the number of users with at least 1Mbit/s minimum end‐to‐end rate increases by 550% when compared with the classical signal‐strength based association. Copyright © 2013 John Wiley & Sons, Ltd.

Wireless mesh network (WMN), radio nodes that form a mesh topology, is an interesting architectural candidate for the future wireless Internet. A dense access network can rapidly be deployed at a reasonable cost because there is no need to wire the mesh access points. However, WMNs typically do not perform well with latency-sensitive data traffic, such as Voice over IP (VoIP), due to restricted bandwidth. In this paper, we present the design, implementation and evaluation of LA-AP-OLSR for hybrid WMNs, where mesh nodes have two radios; a fixed radio for receiving and a switchable radio which rapidly changes channels to send to neighbor nodes. The key novelty of our approach is to use any path routing for latency-sensitive traffic while normal traffic is treated using single path routing. Also, we develop a novel queuing strategy which prioritizes latency-sensitive traffic. A detailed evaluation using KAUMesh, an in-house multi-radio wireless mesh testbed, shows a reduction in latency and packet loss for VoIP traffic without negatively impacting competing TCP background traffic.

Wireless mesh networks (WMNs) consist of several small routers relaying packets wirelessly toward the destination or the Internet. A dense deployment enables the rapid creation of wireless access networks at a reasonable cost. The capacity of a wireless mesh network can be increased by using a diverse set of channels to allow simultaneous transmissions without causing interference. An interesting approach is to use one fixed wireless network interface for receiving and at least one switchable interface for sending. However, such a channel assignment suffers from switching overhead.

Welcome to HotMesh 2013: The Fifth IEEE International Workshop on Hot Topics in Mesh Networking. After the very successful editions in Europe (Lucca, Italy) two years ago and in the U.S. last year, we are happy to host this year the workshop in Europe again.

Channel assignment is among the most challenging issues for multiradio wireless mesh networks, given the variety of objectives that can be pursued and the computational complexity of the resulting problems. The channel assignment problem has been also shown to be interdependent with the routing problem, i.e., the problem to determine the amount of traffic flow to be routed on every link. Such a relationship raises the need to recompute the channel assignment every time the traffic pattern changes. However, channel assignment algorithms designed to assign channels from scratch will likely return a completely different configuration of radios, which would disrupt the network operation for the time required to switch to using the links established on the new channels. As shown by the experiments that we conducted, such a time may not be negligible, due to the resistance of routing protocols designed for wireless ad hoc and mesh networks to rapidly flagging a link as established/lost. Such a consideration, along with the observation that channel assignment algorithms may be suboptimal, led us to the design of a channel reassignment algorithm that takes the current channel assignment into account and attempts to cope with the new traffic pattern in the best manner possible while modifying the channel on a limited number of radios. In this paper, we illustrate such a channel reassignment algorithm and evaluate its performance by means of both simulations and experiments with real hardware.

Wireless hand-off control typically considers only connectivity strength from the mobile terminal to alternative access points. In wireless video streaming, however, where video freezing must be avoided at the mobile terminal, the playout buffer level should also be considered by hand-off control. In this paper, we first develop a model capturing hand-off dynamics under video streaming, and design a new playout buffer aware hand-off control. It aims to avert a video freeze for as long as possible, maximizing the expected time until freezing. We compute the optimal control in the general case of multiple access points and multiple connectivity strength states per channel. The optimal hand-off control is then computationally probed in specific relevant cases. It is demonstrated that there is a certain playout buffer threshold level - a buffer tipping point - above which a hand-off should be attempted. Of course, this tipping point depends on the access point to mobile channel statistics.

We develop an optimization framework for Quality of Experience (QoE)-based routing in multi-service Wireless Mesh Networks (WMNs). The framework takes into account the heterogeneous requirements of different services delivered over a WMN, such that the overall end-user QoE is maximized under given resource constraints. We propose a novel QoE-aware double reinforcement learning strategy for dynamically computing the most efficient routes to deliver the flows of each service type. Comprehensive NS-2-based simulations demonstrate the substantial performance gains that our approach enables over conventional routing techniques such as AODV, with significant improvement over video quality.

poster Share on CloudMAC: torwards software defined WLANs Authors: Jonathan Vestin Karlstad University, Karlstad, Sweden Karlstad University, Karlstad, SwedenView Profile , Peter Dely Karlstad University, Karlstad, Sweden Karlstad University, Karlstad, SwedenView Profile , Andreas Kassler Karlstad University, Karlstad, Sweden Karlstad University, Karlstad, SwedenView Profile , Nico Bayer Telekom Innovation Laboratories, Berlin, Germany Telekom Innovation Laboratories, Berlin, GermanyView Profile , Hans Einsiedler Telekom Innovation Laboratories, Berlin, Germany Telekom Innovation Laboratories, Berlin, GermanyView Profile , Christoph Peylo Telekom Innovation Laboratories, Berlin, Germany Telekom Innovation Laboratories, Berlin, GermanyView Profile Authors Info & Claims Mobicom '12: Proceedings of the 18th annual international conference on Mobile computing and networkingAugust 2012Pages 393–396https://doi.org/10.1145/2348543.2348592Published:22 August 2012Publication History 9citation903DownloadsMetricsTotal Citations9Total Downloads903Last 12 Months4Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

Routing packets over multiple disjoint paths towards a destination can increase network utilization by load-balancing the traffic over the network. In wireless mesh networks, multi-radio multi-channel nodes are often used to create a larger set of interference-free paths thus increasing the chance of load-balancing. The drawback of load-balancing is that different paths might have different delay properties, causing packets to be reordered. This can reduce TCP performance significantly, as reordering is interpreted as a sign of congestion. Packet reordering can be avoided by letting the network layer forward traffic strictly on flow-level. This would avoid the negative drawbacks of packet reordering, but will also limit the ability to achieve optimal network throughput. On the other hand, there are several proposals that try to mitigate the effects of reordering at the transport layer. In this paper, we perform an in-depth evaluation of such TCP reordering mitigations in multi-radio multi-channel wireless mesh networks when using multi-path forwarding. We evaluate two TCP reordering mitigation techniques implemented in the Linux kernel. The transport layer mitigations are compared using different multi-path forwarding strategies. Our findings show that, in general, flow-level forwarding gives the best TCP performance and that transport layer reordering mitigations only marginally can improve performance.

In this paper we study the interactions between peer selection, routing and channel assignment while deploying Peer-to-Peer (P2P) systems over multi-channel multi-radio Wireless Mesh Networks (WMNs). In particular we propose Bestpeer, a novel peer selection algorithm for WMNs which incorporates a cross-layer path load metric and integrates multi-path load balancing capability throughout the P2P download process in order to achieve faster resource disseminations. By comparing Bestpeer with BitTorrent through extensive simulations, we show that we can reduce the time to disseminate a resource by up to 40 %. The key mechanism used by Bestpeer to increase performance in WMNs is the effective exploitation of cross-layer information made available from the routing and channel assignment during the peer selection process in order to balance the load along high capacity paths.

Welcome to HotMesh 2012: The Fourth IEEE International Workshop on Hot Topics in Mesh Networking. After the very successful edition in Europe (Lucca, Italy) last year we are happy to host this year workshop in the U.S.

Routing packets over multiple disjoint paths towards a destination can increase network utilization by load-balancing the traffic over the network. The drawback of load-balancing is that different paths might have different delay properties, causing packets to be reordered. This can reduce TCP performance significantly, as reordering is interpreted as a sign of congestion. Packet reordering can be avoided by letting the network layer route strictly on flow-level. This will, however, also limit the ability to achieve optimal network throughput. There are also several proposals that try to mitigate the effects of reordering at the transport layer. In this paper, we perform an initial evaluation of such TCP reordering mitigations in multi-radio multi-channel wireless mesh networks when using multi-path routing. We evaluate two TCP reordering mitigation techniques implemented in the Linux kernel. The transport layer mitigations are compared using different multi-path routing strategies. Our findings show that, in general, flow-level routing gives the best TCP performance and that transport layer reordering mitigations only marginally can improve performance.

Several protocols for routing and forwarding in Wireless Mesh Networks (WMN) have been proposed, such as AODV, OLSR or B.A.T.M.A.N. However, providing support for e.g. flow-based routing where flows of one source take different paths through the network is hard to implement in a unified way using traditional routing protocols. OpenFlow is an emerging technology which makes network elements such as routers or switches programmable via a standardized interface. By using virtualization and flow-based routing, OpenFlow enables a rapid deployment of novel packet forwarding and routing algorithms, focusing on fixed networks. We propose an architecture that integrates OpenFlow with WMNs and provides such flow-based routing and forwarding capabilities. To demonstrate the feasibility of our OpenFlow based approach, we have implemented a simple solution to solve the problem of client mobility in a WMN which handles the fast migration of client addresses (e.g. IP addresses) between Mesh Access Points and the interaction with re-routing without the need for tunneling. Measurements from a real mesh testbed (KAUMesh) demonstrate the feasibility of our approach based on the evaluation of forwarding performance, control traffic and rule activation time.

Cognitive Radio (CR) technology constitutes a promising approach to increase the capacity of Wireless Mesh Networks (WMNs). Using this technology, Mesh Routers (MRs) and the attached Mesh Clients (MCs) are allowed to opportunis- tically transmit on the licensed band, but under the constraint not to interfere with the Primary Users (PUs) of the spectrum. Thus, the effective deployment of CR- WMNs require that each MR must be able to: sense the current spectrum, select an available PU-free channel and perform the spectrum handoff to a new channel in case of PU arrival on the current one. How to coordinate these actions in the optimal way which maximizes the performance of the CR-WMNs while minimizing the interference to the PUs constitutes an open research issue in CR systems. In this paper, we propose an adaptive spectrum scheduling and allocation scheme which allows a MR to identify the best schedule of (i) when to sense the current channel, (ii) when to transmit, (iii) when to perform a spectrum handoff. Due the large number of parameters involved, we propose Reinforcement Learning (RL) techniques to allow a MR to learn by itself the optimal balance between spectrum sensing-exploitation- exploration actions based on network feedbacks coming from the MCs. We perform extensive simulations which confirm the adaptivity and efficiency of our approach in terms of increased throughput when compared with non-learning based schemes for CR-WMNs.

In multi-radio wireless mesh networks, a network device simultaneously transmits packets over different channels by using multiple radios. Such frequency diversity not only in creases throughput but makes multi-path routing approaches extremely interesting. This is because the channel diversity reduces the risk for intra- and inter-flow interference. A fundamental problem to solve is the forwarding strategy which determines which packets to be sent over what multi-path segments at any given time. Ideally, the forwarding strategy should schedule flows according to the capacity constraints imposed by the channel assignment. However, the possibility to improve MAC layer efficiency by aggregating small packets into larger ones is reduced when packets are forwarded to different next-hops. In this paper, we develop a novel packet forwarding strategy for multi-radio mesh networks that combines the benefits of multi-path routing with packet aggregation. In our cross-layer approach, we effectively trade-off aggregation opportunities with channel diversity. Simulation results show that our approach can improve network throughput and delay by up to 15 percent and 25 percent, respectively, compared with aggregation unaware forwarding strategies.

Urban-X is a new architecture for Multi-Radio Cognitive Mesh Networks based on principles from Dynamic Spectrum Access Networks. In the Urban-X, spectrum sensing and mobility are challenging a transport control protocol (TCP) whose performance is sensitive to varying delay and available bandwidth. Also high packet loss due to primary users lets standard TCP to operate inefficiently. In this study, we propose a new transport protocol TP-UrbanX for the Urban-X which exploits reinforcement based learning techniques to learn an optimal sending rate and adapt this rate given the dynamics of the environment such as spectrum sensing and channel mobility.

ABSTRACT Vehicular communications promise to bring us safer driving and better traffic control. Dedicated short range communications and IEEE 802.11p are now well established standards for vehicular communications. Channels for those systems, however, are of limited capacity and are dedicated to safety applications. Thus, they are not sufficient to support the broad range of services envisioned in VANETs. It is anticipated that vehicles will utilize Wi‐Fi (802.11 a/b/g) to acquire more capacity. Unfortunately, the Wi‐Fi channels in urban areas are already quite heavily subscribed by residential customers. To evade the residential load, in this paper, we propose CoVanet, a cognitive vehicular ad hoc network architecture that allows vehicle radios to access the Wi‐Fi channels ‘opportunistically’, finding least loaded channels. In CoVanet, network topology and channel environment change frequently because of high node mobility. The main contribution of this work is a cognitive ad hoc vehicular routing protocol (CoRoute). CoRoute is a hybrid that utilizes both geographical location and sensed channel information to establish stable, minimum delay routes to the destination. Simulation results show that CoRoute is robust to mobility and to external interference and improves packet delivery ratio on average by 70% with respect to routing without cognitive radios. Copyright © 2011 John Wiley & Sons, Ltd.

Wireless mesh networks (WMNs) are wireless multihop networks comprised of mesh routers, which relay traffic on behalf of clients and other nodes. Using the standard IEEE 802.11 distributed coordination function (DCF) as MAC layer, a node needs to contend for the medium each time it wants to transmit a packet. This creates high overhead in particular for small packets and leads to poor performance for real-time applications such as Voice over IP (VoIP) or online gaming. Burst packet transmission can increase the efficiency. For example, with the Transmission Opportunity limit (TXOPlimit) in IEEE 802.11e, a station may transfer several packets without contending for the channel in between. Similarly, IP packet aggregation combines several IP packets together and sends them as one MAC Service Data Unit. Originally, both schemes have been developed for singlehop networks only. Thus the impact on WMNs is unclear if the packets need to contend over multiple hops. In this paper, we use measurements from a 9-node WMN testbed to compare TXOPs and IP packet aggregation for VoIP in terms of fairness, network capacity and quality of user experience. We show that for low networks loads, both TXOPs and IP packet aggregation increase the VoIP quality compared to IEEE 802.11 DCF. However, in highly loaded networks IP packet aggregation outperforms the other schemes.

A growing trend has emerged in network architecture research to switch focus from Quality of Service (QoS) to Quality of Experience (QoE) optimization. In this paper, we first present QoE models that characterize user satisfaction of video, audio, and data services over wireless networks. We then develop a novel packet scheduling algorithm for multi-hop wireless networks that jointly optimizes the delivery of multiple video, audio, and data flows according to the QoE metrics. We formulate a multidimensional optimization problem that minimizes the overall distortion across all flows for the given network resources on wireless links. Fairness constraints over the flows are also considered as part of the optimization. Our experimental results, obtained with the NS-2 IEEE 802.16 MESH-mode simulator, show that distortion-aware scheduling can significantly increase the perceived quality of different wireless services under bandwidth constraints. Additionally, improved fairness across the competing flows is demonstrated relative to conventional scheduling techniques.

A wireless mesh network (WMN) has been popularly researched as a wireless backbone for Internet access with off-the-shelf and inexpensive equipments. Nowadays, several applications like contents sharing, multicast video streaming, vehicular networks encourage to build mesh networks in urban areas. However, the deployment of WMNs in unlicensed bands of dense urban areas imposes many challenges. While previous research has mostly focused on optimal channel allocation under inter or intra-flow interferences within mesh nodes, the practical deployment of WMNs also requires to consider the interference caused by external entities such as residential access points that do not belong to the WMN. To address this issue, we propose Urban-X which is a first attempt towards a new architecture for Multi-Radio Cognitive Mesh Networks. Based on a cross-layer scheme, we develop novel routing and forwarding schemes reflecting the residential traffic state of external users. Through an extensive simulation analysis using the ns-2 simulator, we demonstrate the feasibility of our routing scheme and we show its robustness to the variations of channel environment and external traffics.

In this paper we address the impact of ACI on dynamic spectrum allocation in multi-radio systems. In particular, we present the benefits of dynamic spectrum allocation by adapting channel bandwidth and distance in order to mitigate ACI in multi-radio mesh networks. Based on our measurement campaign in an indoor mesh testbed, which we extended in order to use adaptive channel width, we evaluate the performance in terms of network throughput. We show that by using channel bandwidth adaptation and antenna separation, the impact of ACI can be significantly reduced. The performed experiments give important insights into ACI in multi-radio cognitive systems which help to develop e.g. better channel assignment algorithms or capacity estimation methods.

Recently, wireless mesh networks (WMNs) have attracted attention as a way to provide alternative Internet connectivity to rural areas or communities. In WMNs, wireless access points communicate with each other wirelessly, forming a true wireless mesh based access network of mesh relay nodes (MRNs). A major problem is, however, scalability of WMNs as well as MAC and PHY layer overhead for packet transmission. Capacity of WMNs can be increased significantly by aggregating (combining) several smaller packets into larger ones. This has previously been demonstrated for voice flows where many packets are of small sizes. TCP could also benefit, as by aggregating several packets together there is a reduced collision risk between TCP DATA and TCP ACKs in addition to the reduced MAC layer contention. However, the traffic pattern when TCP is used is completely different from when voice and UDP flows are used, which sometimes requires different aggregation designs. In this paper, we investigate the impact of packet aggregation on TCP in Wireless Mesh Networks. Using several different scenarios we demonstrate that packet aggregation can not only increase capacity for TCP in such networks but also improve fairness and reduce end-to-end delay.

More and more people use voice over IP instead of traditional phone networks. The Session Initiation Protocol (SIP), a protocol for session management in general, becomes the standard protocol in use. However, SIP relies on central servers. Thus, maintaining the infrastructure is costly. This results in the trend to transfer peer-to-peer (P2P) concepts to SIP to overcome administration efforts due to self-organisation of the participants. In this paper, we propose a novel JXTA-based P2P architecture for SIP signalling. JXTA is an open peer-to-peer platform, which provides basic infrastructure services for building P2P applications. Our concept does not need central entities for maintaining the SIP traffic, while maintaining compatibility with standard SIP. Additionally, we seamlessly added service location capabilities into our infrastructure to support service discovery in dynamic environments. We evaluated our concept in comparison to standard SIP within three scenarios: local area network, point- to-point network and wide area network. The results show that our approach compares well in terms of response time but results in more traffic for maintaining the peer-to-peer overlay.

The deployment of multimedia services in next- generation networks is a challenge due to the high configuration complexity of the streaming process in different stationary and mobile sub-networks and for various user devices. The session initiation protocol (SIP) has proven to be a suitable mechanism to handle the control of multimedia services in such networks. However, the current standardization and implementation of SIP do not allow the simultaneous coordination of multiple concurrent applications on a single device, as the prescribed realization of the SIP state machines (transactions) does not consider mutual access of applications to a single SIP stack. This paper presents a SIP- based mechanism for synchronized management of services in a shared environment. We have developed a middleware that facilitates the uniform access of multiple applications towards one or multiple SIP stacks to enable prioritization of services and centralized resource coordination of concurrent SIP applications on a single terminal or server.

In the field of mobile communications, there coexist various access technologies that support wireless connections, such as Bluetooth, IEEE 802.11 and GPRS. These technologies are diverse in terms of bandwidth, delay, coverage area, etc. It is not uncommon that a mobile device is equipped with multiple wireless network interfaces to achieve efficient and ubiquitous communications. In wireless overlay networks, the mobile node can even utilize multiple network interfaces simultaneously to achieve greater QoS performance. This paper presents the design of such a system that supports multi-homed communications. In particular, we design the policy for mapping different network flows to different interfaces, and propose a load-sharing algorithm based on weighted-round-robin channel selection and jump-ahead packet scheduling. Simulation results show that the algorithm can efficiently distribute data packets among multiple channels to achieve bandwidth aggregation.

Anytime, anyplace application of distributed, multimedia services requires flexible configuration of service quality and coordination of relevant management actions both in mobile terminal and network. This paper presents a new approach for coordinating quality of service (QoS) configuration at multimedia-session establishment and during service adaptations, in accordance with user preferences for service presentation, user contracts with providers, and access/utilization management of the access-provider networks. The general features of an end-to-end negotiation protocol (E2ENP) for multimedia-session management are discussed. E2ENP is designed to minimize signaling load and maximize flexibility of system configuration. The application model of the E2ENP is described and interactions between terminals and access provider are discussed. Some initial measurement results detailing E2ENP performance are presented in consideration with various network configurations.

In this paper we present a system for IP based networks which adapts media streams to the respective characteristics of different (wireless) access networks. Inside the mobile node, an interface selection subsystem monitors characteristics of network interfaces and triggers vertical handovers based on policies supplied by the operator. A service subsystem initiates media stream adaptation to match network resource availability and provider constraints. A media subsystem on mobile and correspondent node transmits the media streams and applies appropriate error correction mechanisms (like FEC, ARQ) depending on network characteristics. We evaluate vertical handover behaviour of our system in a combined GPRS - WLAN testbed and analyze the impact of several error correction mechanisms on the perceived video quality.

Transmitting real-time multimedia streams over heterogeneous mobile networks is a challenging task. Variation in network and system conditions can dramatically affect application performance. When providing end-to-end quality-of-service (QoS) multiple system facets should be coordinated: orchestration of local and peer resources, reservation of network resources, adaptation of multimedia streams, etc. This paper presents an end-to-end negotiation protocol (E2ENP) for negotiating and coordinating QoS on an end-to-end basis both at application and network layer. Based on a flexible extensible markup language (XML) model and extending SDPng concepts, the protocol enables the negotiation of system capabilities and allows provider-services to effectively influence the negotiation process. The aim of the E2ENP design is to optimize the efficiency of multimedia call setup and reduce the time for QoS renegotiations, whenever vertical handovers or spontaneous network reconfigurations occur. The basic protocol is presented, together with implementation and measurement results, stemming from several studies on current and future third-generation/fourth-generation scenarios.

The BRAIN (broadband radio access for IP based networks) project has been proposed to study a system that combines different wireless access network technologies. We present mechanisms to manage end-to-end QoS, including mobile end-systems and IP based wireless access networks (i.e. the BRAIN access network, BAN) in the transmission path. This includes the possible impact of mobility on QoS issues. Supporting QoS in mobile environments is hard to achieve because of potential handovers (horizontal and vertical) and error rate patterns of wireless links in contrast to fixed networks. To solve this issue, QoS signalling and adaptation mechanisms are required. We propose to build upon the BRAIN QoS framework and the BRain ENd Terminal Architecture (BRENTA), which provides the end-system functionality to manage QoS end-to-end from the users and applications point of view. We also show how admission control is embedded in the overall scenario and draft several protocol variants for an end-to-end negotiation of QoS that integrates local resource management and adaptation path strategies derived from users' QoS requirements.

Mobile multimedia systems have to provide mechanisms to cope with the heterogeneity caused by different access network and end system capabilities inherent to wireless communication. Adapting video streams on the fly can be used to match receivers processing and hardware capabilities and the QoS characteristics of wired and wireless links. Within this paper we present a whole suit of filter services that can be applied in order to match frame rate, frame size, visual quality and bandwidth constraints. The filter operate on wavelet coded video streams and adapts them in real-time. A special self learning video filter estimates the source bandwidth of the stream, and times its internal filter operations based on users preferences in order to match a given target bandwidth. We also derive algorithms that map between the filter parameters (i.e. what operations the filters perform) and the resulting bandwidth of the stream.

Next generation IP networks and applications will have to address the increasingly important challenges of wireless access, mobility management, the provision of quality of service (QoS), and multimedia issues. These problems form the basis of the research within the EU financed BRAIN (Broadband Radio Access for IP based Networks) project. The project is developing a novel architecture that will be able to deal with the extreme QoS violations that are likely to occur during a running session that is exposed to the radio access environment. The core of this architecture supports different types of applications. It inherits and develops from the traditional Internet approach, but incorporates aspects of a modern flexible QoS middleware solution. The given problem is addressed in a comprehensive, modular, and open manner, by providing different APIs to different types of applications. It provides powerful functions to application programmers, but does not assume that lower level functionality must be hidden from the application programmer. It encompasses a variety of objects, APIs, end-system mechanisms and protocols to cope with the dynamic variation in mobility management and QoS. This solution will provide applications with more predictable services and allow applications to react in a pre-determined way to QoS violations.

Within this paper we simulate the transmission of MPEG-2 transport stream (TS) packets over a wireless ATM network. Based on a finite state radio channel model for the physical layer of a wireless ATM link including the characteristics of the ATM and MAC layer, different packing schemes are evaluated for encapsulating MPEG-2 transport stream packets in ATM adaptation Layer 5 (AAL5) PDUs. We analyze the performance with respect to both delay and visual quality in terms of PSNR based on a calculated cell error ratio (CER) for each given state of the radio model. The statistics show, that the 1TP (one MPEG-2 TS per AAL5 PDU) scheme outperforms all other packing schemes in terms of visual quality. At medium channel quality (38 dB), quality is judged to be good, although the CER may be as high as 25%.

Distributed multimedia applications with real time capabilities will come to the desktop, as soon as quality of service (QoS) capable networks are easy to program for. Therefore, we propose a QoS application programmer interface (API) for the Java language, which provides an abstraction from QoS capable networks and allows a simple specification of QoS parameters. Platform independence is achieved by Java, which also allows easy integration with the World Wide Web if the classes are used within applets. We first describe our architecture, which abstracts from different QoS capable transport protocols and networks and provides easy to use access to QoS networking. We identify different layers within a distributed multimedia application and show how mapping of QoS parameters at different layers can be accomplished by our API. Different performance evaluations show that our architecture can reach up to 90% the performance of native code measured during a stream test and up to 87% measured during a request response test.

ATM enforces traffic contracts-applications agree to send not more data then negotiated and the network tries to provide predictable service. Thus ATM based applications should be easy to develop. For wireless ATM the situation remains not the same. Hand-off to new a access point may result in invalidating the old contract and dropping connections. Applications have to cope with the new situation by e.g. adapting their bandwidth needs. In the context of this paper, we propose to shift the burden of adaptation from the application to a middleware layer, which also acts as a central co-ordinator for requesting and reserving local resources. Media adaptation is performed by using media filters which are controlled by the QoS-providing middleware. A distributed resource negotiation protocol negotiates end-to-end application QoS parameters and uses QoS mappers to derive the necessary filter settings.