The increasing demand for real-time inference on high-volume network traffic has led to the rise of in-network machine learning, where programmable switches execute various models directly in the data-plane at line rate. Effective network management often involves multiple prediction tasks, such as predicting bit rate, flow size, or traffic class; however, existing solutions deploy separate models for each task, placing a significant burden on the data-plane and leading to substantial resource consumption when deploying multiple tasks. To address this limitation, we introduce MUTA, a novel in-network multi-task learning framework that enables concurrent inference of multiple tasks in the data-plane, without exhausting available resources. MUTA builds a multi-task neural network to share feature representations across tasks and introduces a data-plane mapping methodology to fit it within network switches. Additionally, MUTA enhances scalability by supporting distributed deployment, where different layers of a multi-task model can be offloaded across multiple switches. An orchestrator employs multi-objective optimization to determine optimal model placement in multi-path networks. MUTA is deployed on P4 hardware switches, and is shown to reduce memory requirements by ×10.5, while at the same time improving accuracy by up to 9.14% using limited training data, compared with state-of-the-art single-task learning solutions.

The need for real-time inference of large volumes of data led to the development of in-network machine learning. Programmable network switches can now execute various machine learning models in the data-plane at line rate. While a stream of data may require several prediction tasks, such as predicting bit rate, flow size, or traffic class, current solutions only support separate models for each task. This places a significant burden on the data-plane and leads to substantial resource consumption when deploying multiple tasks. To solve this problem, we introduce MUTA; a novel in-network multi-task learning solution. MUTA enables executing multiple inference tasks concurrently in the data-plane, without exhausting available resources. It introduces a data-plane mapping methodology to fit non-binarized multi-task neural networks within network switches. MUTA is deployed on P4-based hardware switches, and is shown to reduce memory requirements by × 10.5 and improve accuracy by up to 9.14% using limited training data, compared with state-of-the-art single-task learning solutions.

Intelligent data-planes (IDPs) can enhance network service performance and adaptation speed by executing one or more machine learning (ML) models directly on the served flows. The real-time ML inference enables line-speed decision-making for some traffic management functionalities. Due to the inherent scarcity of both the computational and memory resources and the strict high-speed per-packet processing demands, existing IDP deployments either realize only a limited set of ML models such as decision trees, or require substantial modifications in the switch hardware. In this paper, we propose INQ-MLT, a novel ML-based management toolbox to address the aforementioned limitations. INQ-MLT delegates the task of training various ML models to the control-plane. The latter adopts a tailored quantization-aware training process to compensate for the effect of precision loss resulting from quantization. The toolbox then employs a quantization mechanism to transform the trained ML model parameters (e.g., weights and activations) from floating-point representations to compact low-precision fixed integer values that can be easily processed and stored in the data-plane. Finally, the trained model is deployed into the IDP pipeline by restricting all its inference operations to basic arithmetic operations. To analyze the performance of INQ-MLT, we quantify the accuracy loss resulting from the quantization step through rigorous theoretical analysis. A proof-of-concept implementation of the proposed toolbox is developed using P4-based software switches. Experiments on two use-cases demonstrate that the deployed quantized models have almost no loss of accuracy when compared to their floating-point counterparts.

The frequent occurrence of network attacks highlights the criticality of developing effective intrusion detection systems (IDSs) that can promptly detect and respond to malicious flows. The proliferation of programmable devices has opened up new possibilities for integrating intelligent IDSs into the data-plane. This allows the execution of machine learning (ML)-based detection models at line-rate, meeting the low latency requirements of anomaly detection. We propose a two-stage confidence-based Intrusion Detection System (TSCIDS) that aims at early detection while considering the level of certainty of prediction. The control-plane adopts a customized transfer learning scheme, wherein two interdependent convolutional neural network (CNN) models are trained, one using the early context of flows and the other adding the later context. A post-hoc calibration method is applied to improve the performance of models. TSCIDS detects anomalous behavior in different phases of flows while allowing the latter CNN to leverage the hidden state of the early CNN. TSCIDS ensures that the two CNN models are integrated into the data-plane pipeline by building the inference steps of CNN into different modules, using switch-supported operations. Simulation results show that the calibrated model can detect more attacks in the early phase compared to the uncalibrated model. Additionally, the training scheme saves the memory consumption of running models on programmable devices.

Offloading some of the traffic management decision-making functionalities to intelligent data-planes (IDPs) can significantly enhance the accuracy and adaptation speed of network services. An IDP executes, at line-speed, one or more machine learning (ML) models for real-time inference and decision making. Unfortunately, existing IDP deployments either realize only a limited set of ML models such as decision trees or require substantial modifications in the switch hardware. These limitations can be attributed to the inherent scarcity of both the computational and memory resources and the strict high-speed per-packet processing demands. To address the aforementioned limitations, we propose a novel ML-based management framework, the in-network quantized ML architecture (INQ-MLA). First, INQ-MLA delegates the task of training and continuously optimizing the IDP ML model to the control-plane. The latter adopts a tailored quantization-aware training process to compensate for the effect of precision loss due to quantization. Second, INQ-MLA employs an efficient quantization mechanism to transform the trained ML model parameters (e.g., weights and activation functions outputs) from floating-point representations to smaller low precision fixed integer values that can be easily processed and stored in the data-plane. Finally, INQ-MLA ensures that the deployed ML model is integrated into the IDP pipeline by limiting all its execution operations to simplified arithmetic operations that are available in most switches. We developed a proof-of-concept implementation of our proposed architecture using P4-based switches. Experimental results demonstrate that INQ-MLA can achieve a high-level of accuracy at runtime.

Fast and customizable programmable data planes (PDPs) implementing new services such as multi-flow synchronization, on- and in-time delivery, and in-network caching and compression are key enablers to future applications (e.g., streamed holograms, telesurgery, and autonomous industrial systems). This paper outlines the design principles of EP4, an application-aware extended P4-based network architecture that offers hosted applications an extensible catalog of services through its control plane. The latter configures a PDP that can achieve minimal parsing and processing for fast-tracked packets as well as customized processing and forwarding for other packets. An extended parser (eParser) performs the first task, which reduces the necessary latency experienced by packets. Alternatively, adaptive processing is achieved using an enhanced processor (eProcessor) that optionally parses customized headers using just-in-time programmable parsers. It then executes selected P4 packet processing pipelines implementing different services. These programs are installed at runtime without impacting other switch functionalities. Experimental results demonstrate the architecture's enhanced performance compared to current solutions.

A major challenge to the anticipated large scale deployment of virtual network functions (VNFs) at the network edge (NE) is the ability to efficiently allocate and manage its scarce resources to meet these functions workload fluctuations. In this paper, we describe a novel containerized infrastructure manager (cIM) that extends current managers, such as Kubernetes, with the necessary building blocks to provide an accurate resource allocation service to containerized VNFs at scale. The proposed cIM treats the containerized VNF components (cNFCs), as atomic special purpose functions that can be rapidly deployed to form complex network services. The main component of the proposed cIM, the resource reservation manager (RRM), employs concepts of risk pooling in the insurance industry to accurately reserve the needed resources for the hosting containers and meet anticipated cNFCs demand fluctuation. The reserved quota of re-sources ensures the desired availability level of the cNFCs without over-provisioning the scarce resources of the NE. Experimental results demonstrate that our proposed cIM significantly improve the performance of the cNFCs and guarantees their availability with minimal resource reservation.

Managing and orchestrating the behaviour of virtual network functions (VNFs) remains a major challenge due to their heterogeneity and the ever-increasing resource demands of the served flows. In this paper, we propose a novel VNF manager (VNFM) architecture to automate the process of selecting appropriate VNF management actions (e.g., migration and vertical and horizontal scaling) with their corresponding configuration parameters (e.g., migration location or amount of resources needed for scaling). More precisely, we first propose a novel Markov decision process with parameterized actions to accurately describe each VNF and its permissible lifecycle management (LCM) operations. The use of parameterized actions allows us to rigorously represent the functionalities of the VNFM in order perform various operations on the VNFs. Next, we propose a two-stage reinforcement learning (RL) scheme that alternates between learning optimal LCM actions and updating their parameters selection policy. In contrast to existing schemes, the proposed work uniquely provides a holistic management platform that unifies individual efforts targeting single LCM functions such as VNF placement and scaling. Performance evaluation results demonstrate the efficiency of the proposed VNFM in maintaining the required performance level of the VNF while optimizing its resource configurations.

Recent advancements in technology have led to a deluge of big data streams that require real-time analysis with strict latency constraints. A major challenge, however, is determining the amount of resources required by applications processing these streams given their high volume, velocity and variety. The majority of research efforts on resource scaling in the cloud are investigated from the cloud provider's perspective with little consideration for multiple resource bottlenecks. We aim at analyzing the resource scaling problem from an application provider's point of view such that efficient scaling decisions can be made. This paper provides two contributions to the study of resource scaling for big data streaming applications in the cloud. First, we present a Layered Multi-dimensional Hidden Markov Model (LMD-HMM) for managing time-bounded streaming applications. Second, to cater to unbounded streaming applications, we propose a framework based on a Layered Multi-dimensional Hidden Semi-Markov Model (LMD-HSMM). The parameters in our models are evaluated using modified Forward and Backward algorithms. Our detailed experimental evaluation results show that LMD-HMM is very effective with respect to cloud resource prediction for bounded streaming applications running for shorter periods while the LMD-HSMM accurately predicts the resource usage for streaming applications running for longer periods.

Vehicular Ad hoc Networks (VANETs) allow for vehicle-to-vehicle and vehicle-to-infrastructure communications using wireless local area network technologies. The distinctive features of their candidate applications (e.g., collision warning and local traffic information for drivers), resources (e.g., computational sources), and their ability to collect various data from their environment (e.g., vehicular traffic flow patterns) make VANETs a rich resource for information and resources. In this paper, we propose a new methodology to use VANETs to optimize signal control at traffic intersections as well as to create Vehicular Cloud (VC) computing environments. We theoretically analyze the traffic flow patterns in a given road intersection by using the diffusion approximation model. We calculate the probability of clearing the intersection and demonstrate the effect of the traffic patterns on the optimal choice of the traffic signal control parameters. Then, we employ our theoretical analysis to propose a potential solution to construct VANET-assisted VCs. Experimental results verify the correctness of our analysis.

Application offloading to clouds is the key enabler for compute-intensive applications running on mobile devices. An offloading algorithm employs estimated averages of the execution and communication costs of application modules to decide on a modules subset to be offloaded with the objective of minimizing a certain metric (e.g., execution time or energy). This decision is highly affected by the inherent uncertainty arising from the estimated cost averages due to natural fluctuations or measurement inaccuracies. In this article, we propose a novel offloading scheme that takes into consideration these uncertainties. The proposed work first formulates the offloading problem as a tractable robust optimization one where the uncertainty in k cost parameters is incorporated by allowing these parameters to fluctuate within intervals specified from profiling the application and the network. We then show that this problem can be transformed into k + 1 binary linear programs that are solved while preserving the complexity of the original problem. In contrast to existing approaches, the performance of the obtained decision is guaranteed as long as the behavior of the uncertain parameters remains within the given intervals. Performance evaluation results using a face detection and synthetically generated applications with a large number of modules demonstrate the robustness of the obtained offloading decisions.

Named data networking (NDN) is a nascent vision for the future Internet that replaces IP addresses with content names searchable at the network layer. One challenging task for NDN routers is to manage huge forwarding information bases (FIBs) that store next-hop routes to contents. In this article, we propose a family of compressed FIB data structures that significantly reduce the required storage space within the NDN routers. Our first compressed FIB data structure is FCTree. FCTree employs a localized front-coding compression, that eliminates repeated prefixes, to buckets containing partitions of routes. These buckets are then organized in self-balancing trees to speed up the longest prefix match (LPM) operations. We propose two enhancements to FCTree, a statistically compressed FCTree (StFCTree) and a dictionary compressed FCTree (DiFCTree). Both StFCTree and DiFCTree achieve higher compression ratios for NDN FIBs and can be used for FIB updates or exchanges between the forwarding and control planes. Finally, we provide the control plane with several knobs that can be employed to achieve different target trade-offs between the lookup speed and the FIB size in each of these structures. Theoretical analysis along with experimental results demonstrate the significant space savings and performance achieved by the proposed schemes.

Containerization provides a lightweight alternative to the use of virtual machines for potentially reducing service cost and improving cloud resource utilization. A key challenge is how to allocate container resources to multiple competing streaming applications with varying QoS demands running on a heterogeneous cluster of hosts. In this paper, we focus on workload distribution for optimal resource allocation to meet the real-time demands of competing containerized big data streaming applications. We propose a container resource allocation mechanism (CRAM) based on game theory and formulate the problem as an n-player non-cooperative game among a set of heterogeneous containerized streaming applications. From our analysis, we obtain the optimal Nash Equilibrium state where no player can further improve its performance without impairing others. Experimental results demonstrate the effectiveness of our approach, which attempts to equally satisfy each containerized streaming application's request as compared to existing techniques that may treat some applications unfairly.

Named Data Networking (NDN) is a future Internet architecture that replaces IP addresses with namespaces of contents that are searchable at the network layer. A challenging task for NDN routers is to manage forwarding-information bases (FIBs) that store next-hop routes to contents using their stored usually long names or name prefixes. In this paper, we propose FCTree, a compressed FIB data structure that significantly reduces the required storage space at the router and can efficiently meet the demands of having routes that are orders of magnitude larger than IP-based ones in conventional routing tables. FCTree employs a localized front-coding compression to buckets containing partitions of the routes. The top routes in these buckets are then organized in B-ary self-balancing trees. By adjusting the size of the buckets, the router can reach an optimal tradeoff between the latency of the longest prefix matching (LPM) operation and the FIB storage space. In addition, in contrast to existing hash and bloom-filter based solutions, the proposed FCTree structure can significantly reduce the latency required for range and wildcard searches (e.g., for latency sensitive streaming applications or network-layer search engines) where up to k routes are returned if they are prefixed by a requested name. Performance evaluation results demonstrate the significant space savings achieved by FCTree compared to traditional hash-based FIBs.

This work presents a novel statistical cost model for applications that can be offloaded to cloud computing environments. The model constructs a tree structure, referred to as the execution dependency tree (EDT), to accurately represent various execution relations, or dependencies (e.g., sequential, parallel and conditional branching) among the application modules, along its different execution paths. Contrary to existing models that assume fixed average offloading costs, each module's cost is modelled as a random variable described by its Cumulative Distribution Function (CDF) that is statistically estimated through application profiling. Using this model, we generalize the offloading cost optimization functions to those that use more user tailored statistical measures such as cost percentiles. We employ these functions to propose an efficient offloading algorithm based on a dynamic programming formulation. We also show that the proposed model can be used as an efficient tool for application analysis by developers to gain insights on the applications' statistical performance under varying network conditions and users behaviours. Performance evaluation results show that the achieved mean absolute percentage error between the model-based estimated cost and the measured one for the application execution time can be as small as 5 percent for applications with sequential and branching module dependencies.

Recent advancements in technology have led to a deluge of data that require real-time analysis with strict latency constraints. A major challenge, however, is determining the amount of resources required by big data stream processing applications in response to heterogeneous data sources, streaming events, unpredictable data volume and velocity changes. Over-provisioning of resources for peak loads can be wasteful while under-provisioning can have a huge impact on the performance of the streaming applications. The majority of research efforts on resource scaling in the cloud are investigated from the cloud provider's perspective, they focus on web applications and do not consider multiple resource bottlenecks. We aim at analyzing the resource scaling problem from a big data streaming application provider's point of view such that efficient scaling decisions can be made for future resource utilization. This paper proposes a Layered Multi-dimensional Hidden Markov Model (LMD-HMM) for facilitating the management of resource auto-scaling for big data streaming applications in the cloud. Our detailed experimental evaluation shows that LMD-HMM performs best with an accuracy of 98%, outperforming the single-layer hidden markov model.

Cloud clients (CCs) of current distributed cloud applications are still not assured of their service quality, in particular, in terms of the experienced latency. Unfortunately, this is mainly attributed to the unpredictability of the communication links among their hosting distributed data centers. To address this problem, this article introduces a novel virtual-network-as-a-service (VNaaS) model to host these applications. In contrast to existing randomly or statically provisioned inter-data centers bandwidth sharing models, the proposed model allows CCs to accurately express their varying network resources needs, demand constraints and tolerance to the cloud latency. In turn, the model maps these requirements to create inter-data centers virtual links hosting each multiple virtual pipes with differentiated service qualities to carry the CC's various traffic flows. To aid the CCs in optimally determining their VNaaS demands, given the budget constraints of their hosted applications, we also develop a novel demand selection scheme based on a two stage-budget allocation mechanism. In the first budgeting stage, the CC calculates an optimal effective service rate for each of its virtual link along with a corresponding link budget and price index. In the second stage, the virtual link budget is distributed to purchase bandwidth for the link's virtual pipes, each with a given service quality and pricing. We then extend the proposed model to allow the CC to enforce any required virtual links' capacity constraints on the effective service rates resulting from the traffic matrix on the VNaaS. Finally, we develop corresponding differentiated VNaaS pricing and service monitoring mechanisms that can be employed by the cloud service provider (CSP) to regulate the offerings and demands of the distributed cloud services. Performance evaluation results demonstrate the significant improvement in the service quality, the higher utilization of the cloud resources and the increase in the CSP's net profit.

This paper presents a novel distributed Quality of Service (QoS) monitoring scheme for cellular data networks serving highly dynamic users with power- limited devices. The proposed scheme relies on candidate QoS monitoring users that can efficiently submit QoS related measurements on behalf of their neighbors. These candidate users are chosen according to their devices' residual power and transmission capabilities and their estimated remaining service lifetime. Service monitoring users are then selected from these candidates using a novel user-to-user semantic similarity matching algorithm. Simulation results demonstrate the significant gains achieved by the proposed scheme in terms of the reduced traffic overhead and overall consumed users' devices power while achieving a high monitoring accuracy.

Cloud service providers (CSP) strive to effectively provision their cloud resources to ensure that their hosted distributed applications meet their performance guarantees. However, accurately provisioning the inter-data centers network resources remains a challenging problem due to the cloud hosted applications' workload fluctuation. In this paper, we propose a novel approach that enables a CSP to offer Elasticity-as-a-Service (EaaS) for inter-data centers communication in order to guarantee the performance of distributed cloud applications. The contributions of the proposed work are two fold; first, we develop an efficient approach that enables the CSP to estimate and reserve the pool of network resources needed to fulfill the demands imposed by the network workload fluctuations of applications subscribing to this service. The approach allows the CSP to offer communication EaaS at differentiated levels based on the degree of bandwidth-sensitivity of the distributed cloud applications. In order to capture the inter-data centers network activity of hosted applications, we model their workloads using Markovian modeling. The second contribution is a novel dynamic pricing mechanism for network EaaS offerings that can be employed by the CSP to maximize the expected long-term revenue, and to regulate network elastic demands. Performance evaluation results demonstrate the efficiency of our proposed approach, the higher accuracy of our prediction method, and the increase in the CSPs net profit.

This paper presents a new application model and a novel execution path-based algorithm for offloading in Mobile Cloud Computing (MCC) environments. Application offloading is the mechanism by which parts or modules of the application are executed in cloud remote computational services. Such operation saves resources in the mobile device but also incurs costs of accessing the cloud and using the communication network connecting the mobile device and cloud. The offloading problem is to find an offloading decision that minimizes the total cost of executing the application. Previous solutions find a single offloading decision per module. However, the total cost of executing a module also depends on the sequence of module calls leading to its execution. We present a fine grained application model and a fast optimal offloading decision algorithm where multiple offloading decisions are made per module based on the execution paths leading to the module. We evaluate our solution for a face detection mobile applications in multiple network scenarios. We show that our model and algorithm offer offloading decisions that are significant faster than offloading decision by traditional offloading schemes.

This paper presents a novel economic model to regulate capacity sharing in a federation of hybrid cloud providers (CPs). The proposed work models the interactions among the CPs as a repeated game among selfish players that aim at maximizing their profit by selling their unused capacity in the spot market but are uncertain of future workload fluctuations. The proposed work first establishes that the uncertainty in future revenue can act as a participation incentive to sharing in the repeated game. We, then, demonstrate how an efficient sharing strategy can be obtained via solving a simple dynamic programming problem. The obtained strategy is a simple update rule that depends only on the current workloads and a single variable summarizing past interactions. In contrast to existing approaches, the model incorporates historical and expected future revenue as part of the virtual machine (VM) sharing decision. Moreover, these decisions are not enforced neither by a centralized broker nor by predefined agreements. Rather, the proposed model employs a simple grim trigger strategy where a CP is threatened by the elimination of future VM hosting by other CPs. Simulation results demonstrate the performance of the proposed model in terms of the increased profit and the reduction in the variance in the spot market VM availability and prices.

Network virtualization techniques aim at efficiently allocating the underlying substrate network (SN) resources to the hosted virtual networks (VNs). Unfortunately, over time, and due to the frequent initiation and termination of VNs, the available and utilized SN resources become fragmented. This in turn, gradually degrades the performance of these techniques. In this paper, we propose a novel proactive SN resource re-optimization technique that efficiently overcomes the fragmentation problem by performing appropriate re-arrangement, or house cleaning, for the available and utilized SN resources. To minimize the incurred computational overhead, the invocation of this technique is only triggered by certain events such as the departure of an expired VN. The contributions of the proposed work are two fold; first, we develop an efficient technique for the selection and re-allocation of VN portions that are contributing to the fragmentation problem. The technique takes into consideration the trade-off between the benefit from increasing the SN utilization and the cost incurred by the VN migration. The second contribution is novel VN live migration techniques that significantly reduce the service interruption time during migration. Simulation experiments demonstrate the achieved gain in the SN resource utilization as well as in the VN acceptance ratio and the net revenue.

This paper presents a new pricing mechanism for virtual network (VN) services to regulate the demand for their shared substrate network (SN) resources. The contributions of this article are two-fold; first, we introduce a new time-of-use pricing policy for the SN resources that reflects the effect of resource congestion introduced by VN users. The preferences of the VN users are first represented through corresponding demand-utility functions that quantify the sensitivity of the applications hosted by the VNs to resource consumption, time-of-use and prices during peak-demand periods. We then introduce a novel model of time-varying VNs, where users are allowed to up- or down-scale the requested resources to continuously maximize their utility while minimizing the cost of embedding the VNs onto the SN. The second contribution is a novel hierarchical embedding management approach tailored to efficiently map these dynamic VNs. The proposed VN embedding scheme recasts the VN embedding problem as a subgraph matching one, and introduces a simple heuristics-based matching procedure to find a good VN embedding from a number of candidate solutions obtained in parallel. In contrast to existing solutions, the proposed scheme does not impose any limitations on the size or topology of the VN requests. Instead, the search is customized according to the VN size and the associated utility. Experimental results demonstrate the performance achieved by the proposed work in terms of the increased profit, resource utilization and number of accepted requests.

This paper presents a novel distributed scheme for the dynamic construction and management of spontaneous social communities (SSCs) of mobile users in ad hoc networks. A novel model for SSCs that captures their unique dynamic nature, in terms of community structure and interest in different hot-topics over time is first presented. These time-varying interests are represented through a distributively calculated community profile prototype that reflects dominant characteristics of community members. This prototype is then employed to facilitate the identification of new members. Preliminary performance results demonstrate the achieved community stability and content sharing efficiency.

The virtual network (VN) embedding problem is concerned with mapping the nodes and links of a VN request to a shared substrate network while maximizing some objective function such as maximizing profit or resource utilization. This mapping must satisfy specific node capacity and link bandwidth requirements. This paper presents a novel hierarchical approach for scalable VN embedding that achieves a balance between centralized schemes that have a network-wide view of available resources but represent a management bottleneck and scalable distributed ones that incur a high message overhead. The contribution of this work is two fold; we introduce a novel hierarchical substrate management framework that finds more than one candidate VN mapping in parallel, thus, increasing the chances of finding an optimal mapping. The second contribution is a novel VN mapping scheme that recasts the VN mapping problem as a subgraph matching one using modified graph-powers, and introduces a simple heuristic matching scheme to find an efficient VN mapping. In contrast to existing solutions, the proposed framework does not impose any limitations on the size or topology of the VN request, rather the search is tailored based on the VN size. Experimental results demonstrate the performance of the proposed scheme.

This paper presents novel theoretical results for the channel assignment problem in full multi-interface multi-channel wireless networks (fM2WNs). We show that the minimum interference for a fM2WN with n routers each equipped with r radio interfaces can only be achieved with a certain number of channels and derive upper and lower bounds for that number as a function of r. Furthermore, exact values are obtained when certain relations between n and r are satisfied. These bounds are then employed to develop closed-form expressions for the minimum channel interference. Accordingly, a polynomial-time algorithm to find a near-optimal solution is developed. Theoretical bounds and the performance of the developed algorithm are validated through a comparison with exhaustive search results.

In this paper we analyze the characteristics of the solution space of the channel assignment problem in wireless mesh networks where routers are equipped with multiple radio interfaces and can use multiple channels. We show that the solution space is exponentially scaled with respect to the number of communication links, however, high quality solutions lie in dense regions within the solution space. These regions exhibit a high degree of similarity and redundancy. We also analyze the effects of the radio interface constraints on the structure and the size of the solution space. Based on our analysis, we develop a new scheme for channel assignment that dramatically reduces the size of the original solution space into that of a much smaller unconstrained weighted graph coloring problem. This goal is achieved by finding sets of link groups or bindings to represent a good solution structure that meets the radio interface constraints by construction. This structure is then employed to construct a weighted graph coloring problem that is equivalent to the original problem but with a much smaller solution space. Finally, the graph is colored by a heuristic based graph coloring algorithm that takes advantage of the space symmetry to further speed up the assignment process. Experimental results illustrate the superiority of the proposed scheme.

This paper investigates the efficiency of diagnosing network anomalies using concepts of statistical analysis and evidential reasoning. A bi-cycle of auto-regression is first applied to model increments in the values of network monitoring variables to accurately detect network anomalies. To classify the rootcause of the detected anomalies, concepts of evidential reasoning of Dempster-Shafer theory are employed; the root-cause of a network failure is inferred by gathering pieces of evidence concerning different groups of candidate failures obtained from a training set of detected anomalies and their corresponding root-causes. These groups are then refined to infer the exact cause of failure when evidence accumulates using the Dempster rule of combinations. To handle cases of imbalanced training sets, two new approaches for assigning belief values to different anomaly classes are also proposed. Performance analysis and results demonstrate the accuracy of the proposed scheme in detecting anomalies using real data.

This paper describes PACMAN, a policy-based architecture for context management in ambient networks. The proposed architecture is a middleware that interacts with entities responsible for discovering and gathering various context in the execution environment and provides a unified mechanism for the delivery of appropriate contextual information to services and applications. PACMAN utilizes a novel context model based on policy structures to provide a coherent methodology for the representation, dissemination of both simple and complex context. A set of control policies is utilized to realize core management functionalities such as context processing (e.g., aggregation and filtering), storage, access and dissemination. Through the dynamic generation of policies, PACMAN evolves to accommodate new context types as well as emerging applications requirements.

This paper presents a new approach for autonomic management of heterogeneous communication systems by featuring awareness of the surrounding environment's context. The contribution of this work is two-fold; the first is the development of an overall management infrastructure which identifies the necessary components for future autonomic management systems, namely, a context plane, an adaptive policy plane, a logic plane and an event management plane. The second contribution is a new model of context that is based on event calculus and a new methodology for context inference based on circumscriptive reasoning. The model lends itself easily to an automated deduction process on primitive context knowledge to derive more complex context that can be used to efficiently guide the automated management operations.

This paper presents a novel framework for user mobility prediction that can accurately predict the traveling trajectory and destination using knowledge of user's preferences, goals, and analyzed spatial information without imposing any assumptions about the availability of users' movements history. Using concepts of evidential reasoning of Dempster-Shafer's theory, the user's navigation behavior is captured by gathering pieces of evidence concerning different groups of candidate future locations. These groups are then refined to predict the user's future location when evidence accumulate using Dempster rule of combination. Simulation results are presented to demonstrate the performance of the proposed framework.

This paper presents a novel paradigm to approach the issue of autonomous policy-based management of wired/wireless differentiated communication systems. In contrast to existing management approaches which require static a priori policy configurations, policies are created dynamically. The proposed framework addresses the management issue from a new perspective through posing it as a problem of learning from current system behavior, while creating new policies at runtime in response to changing requirements. A hierarchical policy model is used to capture users and administrators' higher level goals into network level objectives. Given sets of network objectives and constraints, policies are assembled at runtime. The new approach gives more flexibility to users and applications to dynamically change their quality-of-service (QoS) requirements while maintaining a smooth delivery of QoS through network monitors feedback. Simulation results demonstrate the performance of the proposed work.

User Mobility prediction represents a key component in assisting handoff management, resource reservation, and service preconfiguration. However, most of the existing approaches presume that the user travels in an a priori known pattern with some regularity; an assumption that may not always hold. This paper presents a novel framework for user mobility prediction that can accurately predict the traveling trajectory and destination using knowledge of user's preferences, goals, and analyzed spatial information without imposing any assumptions about the availability of users' movements history. This framework thus incorporates the notion of combining user context and spatial conceptual maps in the prediction process. The main objective of this notion is to circumvent the difficulties that arise in predicting the user's future location when adequate knowledge about the history of user's traveling patterns is not available. Using concepts of evidential reasoning of Dempster-Shafer's theory, the user's navigation behavior is captured by gathering pieces of evidence concerning different groups of candidate future locations. These groups are then refined to predict the user's future location when evidence accumulates using the Dempster rule of combination. Simulation results are presented to demonstrate the performance of the proposed framework.