The advancement of big data has paved the way for the development of intelligent and smart applications; however, privacy concerns often hinder fully realizing their benefits. Federated Learning (FL) has emerged as a promising framework for enhancing privacy while training models collaboratively across decentralized data sources. However, it remains susceptible to poisoning attacks, severely undermining its effectiveness. Existing robust aggregation techniques often struggle with the sensitivity of data distributions, and cluster-based strategies often fail to cluster poisoned model updates correctly. The direction obtained from the signs of the gradient mostly solves these problems but remains vulnerable to dying ReLU problems and usually becomes sensitive to outliers. In this paper, we introduce SignDefence, a sign direction and LeakyReLU-based aggregation technique, which considers the direction of the gradients and overcomes the performance issues related to the dying ReLU problem. Moreover, the proposed SignDefence computes Jaccard Similarity over binary encoded model weights and remains robust across sparse data. The experimental results suggest that the proposed technique shows consistently better accuracy and F1 score than the state-of-the-art techniques, without attack and under different attack scenarios.

Integrating accurate resource usage prediction with cloud management systems is critical to optimize resource utilization and improve operational efficiency. The dynamic and non-linear usage patterns of resources in cloud data centers pose significant challenges for predictive modeling. Traditional single-output models, designed to predict a single value, often struggle to capture the complexities of series resource usage patterns. Current predictive models do not consider the interdependencies and interactions of various resource usage patterns in a sequence. Therefore, it is necessary to develop more robust predictive methods that can predict a series of resource usage patterns. This study introduces an innovative predictive model that uses multi-output regression combined with time series windowing, usage pattern clustering, and various transformation methods to predict a series of resource usage with high precision for heterogeneous cloud computing systems. Transformation methods include a power transformer to normalize the data distribution, a standard scaler to standardize the feature space, a polynomial transformation to enhance model complexity, and principal component analysis to reduce the dimensionality of the training feature space. The proposed model is evaluated using multi-output regression benchmarks with real cloud workloads and various evaluation metrics. The results demonstrated that the proposed approach significantly improves prediction accuracy while reducing training costs, offering substantial potential for improved performance and efficiency in cloud computing operations.

Large Language Models (LLMs) have advanced significantly in Natural Language Processing (NLP) over the past few years. Ongoing research continues exploring their capabilities in recommendation systems, aiming to enhance user-tailored content delivery efficiency, accuracy, and personalisation. The investigation introduces a novel approach to integration possibilities of open-source Language Model (LLM) technology—FLAN-T5, Falcon, Vicuna, UL2, and LLAMA—into anime recommendation systems. The research delves into creating personalised recommendations by inputting anime titles, genres, and descriptions into these LLMs. Furthermore, it harnesses LLMs to explain these recommendations, bolstering user engagement and amplifying transparency in the recommendation process. The findings clearly show that using open-source LLMs for anime recommendations works well. It proves that these techniques have great potential to make anime suggestions better.

Load prediction is a critical component of effective resource management in cloud computing. It ensures optimal performance and efficiency by anticipating overload, underload, and normal load periods. However, achieving accurate predictions remains challenging due to the highly dynamic and often non-linear workload patterns typical in cloud environments. Traditional methods, while helpful, have shown limitations in handling these complexities. Machine learning techniques, specifically imbalance and ensemble learning, have shown potential for improving prediction accuracy. Imbalance learning addresses the uneven distribution of load states, while ensemble learning combines multiple models to achieve better predictive performance. It is possible to create a more robust and accurate load prediction system by leveraging these two methods. This paper explores the application of imbalance and ensemble learning to improve load prediction in cloud computing systems. Through an experimental study, we illustrate how these techniques outperform traditional methods, offering potential improvements to the performance and efficiency of cloud computing operations.

Federated Learning (FL) is an emerging paradigm that enables multiple clients to train a global model collaboratively without sharing their privacy-sensitive data. However, one of the significant challenges in FL is the aggregation of the model updates from different client devices, as malicious participants acting as Byzantine attackers can craft the model update and poison the global model. The state-of-the-art defence mechanisms mostly rely on aggregation-based security defences to improve the degraded accuracy. However, preventing attacker's participation in the training can have an impact on improving the global model's accuracy. Therefore, in this paper, FedChallenger, a dual-layer defence mechanism, is proposed, which attempts to detect and prevent malicious participation in the FL training process in its first layer. The other layer incorporates a trimmed-mean aggregation strategy, where pairwise cosine similarity identifies malicious updates and removes entire client updates from federated averaging. Extensive experiments using the BloodMNIST dataset validate that the FedChallenger gains nearly 85%, 80%, 15%, and 4% accuracy with more than 1.2 times faster convergence rate over the state-of-the-art Byzantine resilient aggregation strategies called FedAvg, Fang, Krum, and Trimmed-Mean approach, respectively, on 40% compromised devices. Above all, it shows consistently better results than them in both attack and non-attack scenarios.

The Infrastructure as a Service (IaaS) cloud industry that relies on leasing virtual machines (VMs) has significant portion of business values of finding the dynamic equilibrium between two conflicting phenomena: underutilization and surging congestion. Spot instance has been proposed as an elegant solution to overcome these challenges, with the ultimate goal to achieve greater profits. However, previous studies on recent spot pricing schemes reveal artificial pricing policies that do not comply with the dynamic nature of these phenomena. Motivated by these facts, this paper investigates dynamic pricing of stagnant resources in order to maximize cloud revenue. Specifically, our proposed approach manages multiple classes of virtual machines in order to achieve the maximum expected revenue within a finite discrete time horizon. For this sake, the proposed approach leverages the Markov decision processes with a number of properties under optimum controlling conditions that characterize a model's behaviour. Further, this approach applies approximate stochastic dynamic programming using linear programming to create a practical model. Experimental results confirm that this approach of dynamic pricing can scale up or down the price efficiently and effectively, according to the stagnant resources and the load thresholds. These results provide significant insights to maximizing the IaaS cloud revenue.

Network Function Virtualization (NFV) is a growing computing paradigm for rapid and economical provisioning of networking services (NSs). In NFV, NS is provided through a set of Virtual Network Functions (VNFs) that are hosted on the underlying infrastructure offered by the service provider. Quality of the NS such as availability, as well as the overall cost of providing the NS in NFV domain are raised as concern issues especially their objectives go against each other. This paper tackles two main problems that are directly related to the cost and availability of the NS, VNFs configuration types selection problem and VNF placement problem. Therefore, we propose two Mixed Integer Linear Programming (MILP) optimization models to address both problems and find solutions. We build a proof of concept to evaluate our proposed solutions and compare them with existing solutions from the literature. The results show that our proposed solutions can reduce the overall cost of requested NSs without violating their availability requirements.

Elastic cloud computing model rely on clear definition of workload demand capacity size and cloud resources provision units. These two factors are unknown for any running cloud model, because of the dynamic changes of workload and cloud data center provisioning resources reconfiguration characteristics. These can be defined as unlabeled data. To achieve an accurate elastic scaling, unlabeled data set should be marked and labeled to finite set of workload demand classes and provisioned resources classes. This work introduces a multiple criteria attribute, k-means clustering, for cloud data center elastic model to achieve a commensurate mapping between workload class and provisioned class. Two validation methods for k-means clustering have been applied to validate the cluster group sets, obtaining a good and reasonable mapping for demand and provisioned classes with accepted time and space complexity. Two groups of sets have been generated for workload demands and for resources provisioned, and a simple look-up mapping has been applied using set joint theory.

In this paper the technology of collecting the logs and the logs message format is considered in addition to investigating multiple log data sources as an input to the cloud management system. A comparison between message exchange technologies (JSON, XML) is evaluated with the latest message format technology (Google Protocol Buffer) when used in combination with the message transmission protocols (XML-RPC, REST, Network Socket). In addition, the sampling rate that gives the accurate reading of resource usage is investigated, which is used to select among different log data sources to achieve the accurate log update time. Logs sampling rate of 1.0 second is found to be the best with "xentop" data source. The result of the experiment shows using Protocol Buffer with Socket protocol gives the best results in reducing message size. Network socket with JSON gives the best processing delay and traveling time.

Cloud is a popular and attractive paradigm for providing online computing services to the end users. Recently many of the users move their business applications to the cloud and become tenants for the cloud service providers. Some tenants expect their applications that are provided as services to be highly available (HA) at any time. Managing HA of applications services in the cloud is a big challenge due to the dynamic nature and the huge number of the provision services in the cloud. Limited number of solutions address the HA of services in the cloud platforms that use containers instead of Virtual Machines (VMs). In addition, HA measurements are still missing by the proposed solutions in the literature. Therefore, in this article we propose a framework to incorporate the HA feature for the applications that are deployed in cloud platforms that use the containers. The framework depends on the novel idea of integrating HA middlewares OpenSAF and Pacemaker with the containers to manage HA of the applications services. As a proof of concept, we build a prototype for our framework using our private cloud Container-based platform. For the evaluation purposes, we compare the same framework using our cloud VM-based platform. The measurements show the ability of the proposed framework to manage HA of different services using the containers with faster service recovery time and shorter service outage time than using the VMs.

Cloud computing model is the most complex computing model that requires implementing effective techniques to manage infrastructure resources of datacenters. Unproductive tasks scheduling can lead to an increase in the operational cost of cloud provider side, which in turn increases the cloud services cost at cloud consumer side. One of the effective techniques to address these issues in cloud datacenters is the elasticity by allowing dynamic resource provisioning based on the current demand and varying workload running upon virtual machines (VMs) over time. This leads to an increase in the resource utilization, and reduced power consumption by turning off the idle physical machines. However, the dynamic resource provisioning due to the growing service demand and higher quality of service requirements of the users can cause a violation of service level agreement. In this paper, we propose a model based on linear regression to manage and reformulate cloud users requests and dynamically generating rules based on historical data of their requests in order to update association functions to address and adapt the changes of different types of workloads running on the cloud provider datacenter. The experiments and simulation results based on dynamic workloads show the proposed algorithm significantly increases the resource utilization on cloud datacenter.

Cloud is a popular paradigm for providing online computing services to the end users. Recently many of the cloud service providers use containers instead of Virtual Machines (VMs) to host the applications. Using containers raises the application service availability concerns. Availability is a non-functional requirement that refers to the percentage of time the service is available for the end user. Scheduling containers (host application) on physical/virtual hosts has direct impact on the availability of the application service that is provided to the end user. According to the best of our knowledge, the existing container scheduling solutions do not directly address the availability of the application service. In this article we propose a new Availability-Aware container scheduling strategy that aims to increase the availability level of the application service in the cloud container-based platform. The strategy selects VMs and hosts that have higher availability values within constraints to schedule the containers in efficient way. We compare the proposed strategy with other container scheduling strategies that are used by Docker container platform. The results shown that the Availability-Aware strategy achieves higher service availability levels, and acceptable physical host CPU utilization.

Live VM migration is a technique that consists of a selection process and a migration process to migrate a VM from one host to another in the same data center without changing the IP address, or in a different data center with necessity to the VM to get a new IP address. The changing of IP address results into a mobility problem, which may render the service unreachable. In this paper, we propose a system model that selects data center randomly for VM placement while reducing this IP address reconfiguration. A new metric is proposed to indicate number of users that need IP reconfiguration. We extended CloudSim to simulate our work to identify the number of IP reconfigurations required for VM migration across the data centers on random workload.

Heterogeneous wireless access network (HWAN), an integration of different radio access technologies (RATs) in an overlapping zone, supports bandwidth hungry application and fulfills the demands for high data rates. In this paper, we explored a novel hybrid scheme for RAT selection in HWAN, a two step process, where both a central controller node (CCN) and user device (UD) are involved in the process of network selection. During the first step UD screens the available list of scanned networks based on received signal strength and user mobility profile. The results for the first step of RAT screening using multiplicative exponential weighting method (MEW) are compared with multi criteria simple additive weighting (SAW) utility function. In our second step the CCN takes multi criteria related to application, terminal and network, and generates a sorted list of the most appropriate RATs based on evaluating MEW utility function. The CCN, then associates users to one (single connection) or more available RATs (multi-homed). Using Matlab based simulations, the process of RATs ranking and association is elaborated by calculating final utilities of different networks. The impact of different crucial criteria on RATs ranking results have been explored. Furthermore, we compared our proposed hybrid approach with the traditional mechanisms. The simulation results show that the decision of our proposed hybrid mechanism is more precise than the existing traditional approaches.

Spectrum sensing is the first step to overcome the spectrum scarcity problem in Cognitive Radio Networks (CRNs) wherein all unutilized subbands in the radio environment are explored for better spectrum utilization. Adversary nodes can threaten these spectrum sensing results by launching passive and active attacks that prevent legitimate nodes from using the spectrum efficiently. Securing the spectrum sensing process has become an important issue in CRNs in order to ensure reliable and secure spectrum sensing and fair management of resources. In this paper, a novel collaborative approach during spectrum sensing process is proposed. It monitors the behavior of sensing nodes and identifies the malicious and misbehaving sensing nodes. The proposed approach measures the node’s sensing reliability using a value called belief level. All the sensing nodes are grouped into a specific number of clusters. In each cluster, a sensing node is selected as a cluster head that is responsible for collecting sensing-reputation reports from different cognitive nodes about each node in the same cluster. The cluster head analyzes information to monitor and judge the nodes’ behavior. By simulating the proposed approach, we showed its importance and its efficiency for achieving better spectrum security by mitigating multiple passive and active attacks.

Live virtual machine migration is an essential tool for dynamic resource management in current data centers. Many techniques have been developed to achieve this goal with minimum service interruption. In this paper, we proposea pre-copy live VM migration using Distributed SharedMemory (DSM) computing model. The setup is built usingtwo identical computation nodes to construct the environmentservices architecture namely the virtualization infrastructure, the shared storage server, and the DSM and High PerformanceComputing (HPC) cluster. The custom DSM framework isbased on a low latency memory update Grappa. HPC clusterwith OPENMPI and MPI libraries support parallelization andauto-parallelization work load by using CPUs computationnodes. The DSM allows the cluster CPUs to access the samememory space pages resulting in a lower memory data updatesbased on locality attributes updates, which reduces the amountof data transferred through the network. This model achieves agood enhancement of the live VM migration metrics. Downtimeis reduced by 50% in the idle workload of Windows VM and66.6% in case of Ubuntu Linux idle workload. In general, thismodel not only reduces the downtime and the total amountof data sent, but also does not degrade other metrics like thetotal migration time and the application performance.

Routing in cognitive radio networks (CRNs) faces many limitations that make it challenging. First, traditional routing protocols cannot be directly applied in CRNs because they consider fixed frequency band. Second, cognitive radio enables dynamic spectrum access which causes adverse effects on network performance. Third, effective routing in CR Networks (CRNs) needs local and continual knowledge of its environment. Last, presence of malicious nodes and their misbehaving activities affect the route establishment and therefore reduce the network performance. In this paper, we address such limitations by combining spectrum sensing and routing to propose a novel routing algorithm that uses nodes' behavior during spectrum sensing phase as a routing metric. Through the spectrum sensing phase, nodes behavior is measured through a parameter called belief level (BL), which describes the node's reliability to correctly sense the spectrum and to use spectrum channels accordingly. Moreover, we secure the routing requests and reply messages by encrypting them utilizing the existing cryptography techniques. The proposed approach is designed to maximize security level of paths, minimize the effects of licensed users activity over spectrum channels, and reduce the total channels cost over the best path(s). Evaluation of the proposed approach shows that its performance outperforms many current state-of-the-art routing algorithms used in CRNs in terms of end-to-end delay, packet delivery ratio, and packet loss ratio.

Cloud computing providers in the infrastructure as a service (IaaS) layer provide their utility computing and IT services as virtual machines to customers, who then pay for resources based on time usage. One of the most subtle challenges is pricing stagnant resources dynamically, which combines the static pricing strategy of active resources to maximize cloud computing profits. This paper investigates cloud dynamic pricing and proposes an efficient model that manages virtual machines in regards to revenue management, formulating the maximum expected reward under discrete finite horizon Markovian decisions, characterizing model properties under optimum controlling conditions, approximating optimal dynamic programming policy using a linear programming approach, developing a new algorithm based on this approximation, and finally presenting evaluation results. Our results provide fundamental insights into cloud computing revenue.

Due to the increased use of sensor nodes in a variety of application fields, wireless sensor networks need to handle heterogeneous traffic with diverse priorities to achieve the required quality of service. In this paper, we address the cross layer quality of service-aware scheduling for wireless sensor network with respect to delay and reliability in an energy efficient way. A node disjoint multipath routing is used and a QoS-aware priority scheduling considering MAC layer is proposed to ensure that real time and non-real time traffic achieve their desired QoS while alleviating congestion in the network. We evaluate our algorithm with extensive simulations and the results have demonstrated the effectiveness of our proposed scheme for different metrics.

Recently, several analytical models have been proposed to evaluate the performance of IEEE 802.11 RTS/CTS access mode in non-saturation traffic conditions. However, to the best of our knowledge, none of these models take into account the IEEE 802.11 recovery mechanism for control and data packets together in non-saturation mode and in a presence of transmission errors. In this paper, we propose an M/G/1/K queue with independent samples from saturation analysis to provide a new insight into the QoS performance and queueing behavior of the IEEE 802.11 system. Extensive simulation and analysis results show that our analytical model can accurately predict the throughput, QoS, and queueing characteristics of the RTS/CTS under different channel and traffic conditions.

IEEE 802.11 network uses the physical carrier sensing and RTS/CTS handshake as the main two techniques to combat interference and hidden node problem. But both techniques do not function well if the interferer or hidden node is beyond the transmission range of receivers. In our model, we take this fact into consideration when analyzing the performance of IEEE 802.11 in multi-hop wireless network. Instead of analyzing the network based on the behavior of the transmitter node, we take the receiver node as the point of reference. Mainly, we divide the network around the multi-hop path into a congregation of interleaved single hop subnetworks, i.e., each subnetwork covers up to 2-hop neighbors of one of the reference nodes. We use a finite capacity M/G/1/K queue and independent samples from saturation analysis to analyze the nonsaturation performance of the single hop wireless network. Then, we present a general analytical model and an iterative method to analyze the performance of the multi-hop path. Simulation results show good agreement with the analytical results.

Mobile devices such as laptops and smartphones are becoming increasingly ubiquitous in metropolitan environments which contain numerous heterogeneous networks offering high bandwidth at low cost. As a result, content providers are streaming significant amounts of video content and latency sensitive data streams such as VoIP. In order for the mobile user to maintain a high quality of service, a vertical handover (VHO) mechanism is required to efficiently facilitate the exchange between heterogeneous networks in a seamless manner. Grey system theory has been used in a wide range of systems including economic, financial, transportation, and military to accurately forecast time series based on limited information. In this paper we build on a novel reputation based VHO decision rating system by proposing the use of the grey model first order one variable, GM(1,1), in the handover decision making progress. The low complexity of the GM(1,1) model allows for a quick and efficient prediction of the future reputation score for a given network, providing deeper insight into the current state of the target network.

Maintaining high performance in parallel processing routers while preserving packet ordering within the flows is a difficult problem. To preserve packet ordering, hashing at the flow level has been used to distributed packet processing workload among the router processing units. Even though it preserves ordering, hashing alone may cause significant workload imbalance and thus adaptive methods are usually needed. In this paper, we present an input port selection scheme that can be augmented with the adaptive Highest Random Weight (adaptive HRW) method. The adaptive HRW is a hash-based method that works at the flow level and is used to balance packet processing workload among the router processing units. When imbalance occurs, the adaptive HRW method triggers all input ports to re-balance their workload among the processing units. When augmented the selection scheme, the adaptive HRW method should be able to identify the subset of input ports responsible for the imbalance. The simulation results show that deploying the selection scheme with the adaptive HRW significantly reduces the number of flows remapped while balancing the packet processing workload among the router processing units.

A novel routing architecture that balances incoming Internet flows over the Valiant Load-Balanced (VLB) networks is proposed. The architecture is based on the adaptive highest random weight (Adaptive HRW) algorithm proposed to design load balanced Internet routers. To reduce flow remapping, the architecture extends the adaptive HRW algorithm with a minimal flow remapping selection scheme that identifies the traffic causing imbalance in the network and needs to be rerouted, where rerouting is implemented by adapting the weight vectors associated with the traffic. Compared to the adaptive HRW method, the selection scheme further reduces flow remapping and the effect of packets reordering. The architecture is stateless and can compute routes quickly based on the packet flow identifier. This is important when deploying the VLB network as a backbone network where the number of flows is large and storing flow state information in lookup tables could limit the network performance.

Existing 2-D Markov chain models of IEEE 802.11 DCF mechanism are not capable to model the performance of the RTS/CTS access mode with finite buffer in imperfect channels due to the lack of adequate buffer and data retransmission limit models. This paper presents a discrete-time 4-D Markov chain model that integrates in addition to the data and control retransmission limits, the finite load, finite buffer capacity, and quality of the received data into a one model. Specifically, the additional two dimensions model the data retransmission limit and buffer capacity. Moreover, a single state, the idle state, is added to model the unsaturated condition. The 4-D model provides a new insight into QoS performance and queueing behavior of the IEEE 802.11 system. Simulation results also indicate that the analytical analysis is fairly accurate.

This paper introduces an accurate analysis using three dimensional Markov chain modeling to compute the IEEE 802.11 DCF performance under heavy traffic conditions and absence of hidden terminals. The proposed model matches the real implementation of the DCF as presented in the standard through considering the impact of retry limits of control and data frames jointly on the performance of DCF mechanism. Moreover, transmission errors are added to the model as constant frame error probabilities. In addition to the throughout efficiency, this analytical analysis calculates the average packet delay, the packet drop probability and the average packet drop time for the DCF access modes, basic and RTS/CTS. We show that our proposed model is a more general model compared to the other models that are presented in the literature, which leads to more accurate performance analysis. Simulation results validate the accuracy of our analytical analysis. Moreover, we prove the generality and validate the correctness of our analysis by showing that other models appeared in literature are special cases from our proposed model. Moreover, the impact of the retry limits and the network size on the performance of IEEE 802.11 DCF is presented.

Wireless mesh networks (WMNs) have emerged recently to extend Internet access and other networking services. WMNs routers provide network access to the clients and other networking functions such as routing, and packet forwarding. Bandwidth scarcity is the main challenge that limits the performance of WMNs. Although considerable research has been conducted on spectrum allocation, spectrum management is still considered an important open problem. This problem can be solved using cognitive radio technology that allows radios to intelligently locate free frequencies and use them efficiently. In this work, we propose a new spectrum management scheme that supports local and global management for a wireless network. Our scheme is based on clusters where the coordinator for each cluster manages spectrum information by keeping the required information at cluster level and for the whole network. Our scheme provides robust operation against any cluster head failure, as well as clients mobility.

Abstract Efficient utilization of bandwidth and high data rates have a great impact on the performance of wireless networks. The cdma2000 1xEV‐DO standard provides high‐speed wireless data services to mobile subscribers based on CDMA technology. In this paper, we study the bandwidth utilization for the 1xEV‐DO packet mode standard. In particular, we develop an analytical model for lowest‐rate‐first, highest‐rate‐first priority scheduling techniques, and two round‐robin fair scheduling techniques over the reverse data channel in cdma2000 1xEV‐DO. For these four scheduling techniques, the distribution of the mobile stations (MSs) among the possible data rates is modeled as a Markov process. An analytical expression for the steady state system throughput is derived from the steady state distribution of the above Markov process. The developed model is validated through simulations. Copyright © 2008 John Wiley & Sons, Ltd.

A new MPLS-based recovery approach for multicast trees is proposed. The main objective of the proposed approach is to trade off the extreme capacity consumed in the local recovery approach that builds a backup path for each element in the tree and the extreme time it takes to recover from the failure in the end-to-end recovery approach. Although the presented approach can be implemented in any large backbone network that employs multicast communication mode, we concentrate our discussion on the MPLS networks. After dividing the multicast tree into several domains, backup paths are set up between the border routers in each domain. In terms of the total reserved capacity, simulation results have shown that the performance of the proposed approach is close to the one produced by global recovery approach. In addition, the proposed approach outperforms the global recovery approach in terms of the average time needed to recover from link/node failure.

The emerging IEEE 802.11e standard is expected to provide service differentiation and resource allocation for various types of real-time traffic. To support the transmission of voice and multimedia data with performance guarantees, it is crucial to design efficient algorithms for admission control and resource allocation. Several methods have been proposed. However, most of these proposed methods may not be efficient because they assign channel access parameters (CAPs) according to the access category (AC) a flow is mapped into rather than based on the absolute QoS requirements of the flow. Using simulations we highlight the shortcomings of current admission control methods and accordingly we propose a flow-based service differentiation mechanism, which select Channel Access Parameters (CAPs) based on each traffic QoS requirements.

We propose a two-step method to enhance the dual-failure restorability in p-cycle protected mesh networks that are optimally designed to withstand only single failures. Our two-step method relies on finding alternate routes to repair the non functional p-cycles upon the first failure and protect the exposed spans. We then compare our proposed method with the cycle reconfiguration method proposed in [9] in which the spare capacity is reconfigured dynamically (either incrementally or completely) after the first failure. We find out the additional spare capacity required for dual failure restorability as compared to single failure restorability for our proposed method as well as the incremental and complete cycle reconfiguration methods. We conclude our findings based on our results and discuss the advantages and disadvantages of each of these methods.

The capacity demands required to handle new wireless services such as e-mail, web surfing, and video streaming led to a new era of wireless technologies. cdma2000 1xEV-DV is considered to be the solution for the ever-growing demand of high-speed wireless packet data transmission. We investigate several aspects related to autonomous rate assignment schemes over the cdma2000 1xEV-DV reverse packet data channel. In particular, based on the developed rise over thermal model, we provide an upper bound for the reverse packet data channel throughput as a function of the number of mobile stations that are allowed to transmit instantaneously on each time slot. We also propose several autonomous rate assignment schemes that provide a significant throughput improvement relative to other published schemes.

The emerging IEEE 802.11e standard for wireless local area networks (WLANs) has been proposed to support quality of service (QoS) by assigning different channel access parameters (CAPs) to different access categories (ACs). As an important part of QoS, an admission control scheme is required to maximally utilize the wireless medium resources and to efficiently admit the upcoming real time traffic while not compromising the QoS of existing traffic. In this paper, we propose a novel admission and congestion control scheme which obtains the admission control parameters through existing analytical model and traffic QoS requirements. It then dynamically updates the CAPs based on periodical monitoring of current channel conditions. Through numerical analysis and extensive simulation, results show that such a scheme could provide the guaranteed QoS for admitted real-time traffic in terms of guaranteed throughput achievement, bounded maximum delay and bounded maximum dropping rate while maintaining good channel utilization

We study the unavailability of end-to-end traffic in p-cycle based mesh networks, which are designed to protect against single link failures. It has been shown earlier by Grover and Clouqueur that the p-cycle length as well as its topology play a vital role in determining the availability of span(s) which are protected by the p-cycle. Similarly, we derive the relationship between the unavailability of a span(s) and the topology of the p-cycles) which is allocated for the restoration of the span(s). Based on these insights and on the fact that the end-to-end unavailability of a working path depends not only on the length of the restoration path but also on the number of spans along the working path, we try to design a method for allocating p-cycles such that the end-to-end unavailability is bounded by an upper limit and the upper limit can be varied as desired. As expected, results show that more capacity is required to guarantee a lower end-to-end unavailability. Our results also show that shorter service paths tend to use longer p-cycles than longer service paths, to obtain the same level of availability; this is expected since the path length, apart from the p-cycle length, also plays a role in determining the availability of the service path. We compare this formulation with a formulation which rather limits the hop count of candidate p-cycles to provide a lower end-to-end unavailability. We notice that directly limiting the end-to-end unavailability, as proposed by this paper, gives better results in terms of spare capacity redundancy than limiting the hop count of p-cycles. That is because the former allows shorter working paths to use p-cycles with higher hop count and therefore a better utilization of the allocated spare capacity

A new scheme for deploying multicasting in MPLS networks is proposed. Each possible tree in an MPLS network is assigned a number, which is then used to classify the corresponding multicast session into its FEC. We call this approach tree numbering (TN). It provides the capability to aggregate different multicast flows (sessions) having the same tree "shape". The assigned number is calculated distributedly by adding the partial weight values generated by the LSRs and the ingress LER of the corresponding tree. The key point in our approach is that the assigned numbers needed to distinguish the "shapes" of all the possible trees depend on the number of possible egress LERs could be reached by that ingress LERs and not on the number of the LSRs (core routers). In terms of the memory size, the proposed approach outperforms the approach that stores the IP addresses of the multicast tree or the one that store the concatenation of the sub codes generated by the ingress LER and the LSRs.

Mobile ad-hoc network (MANET) is an active topic of research for its potential of providing pervasive services anywhere and anytime, even though some challenges need to be handled first. In this paper, we propose an overlay multicast framework to handle multicasts in MANET environment in a flexible way. Our approach is using SIP to discover peers to set up a meshed overlay network first, then overlay multicast trees are set up on demand. To cope with the bandwidth limitation problem, the meshed clustering structure is adopted, so that the virtual meshed network topology gradually adopts the changes in the underlying network in a distributed manner and the multicast tree is adapted to the updated topology information accordingly.

Discusses fault tolerance in computer communication protocols, modeled by communicating finite state machines, by providing an efficient algorithmic procedure for recovery in such systems. Even when the communication network is reliable and maintains the order of messages, any kind of transient error that may not be detected immediately could contaminate the system, resulting in protocol failure. To achieve fault-tolerance, the protocol must be able to detect the error, and then it must recover from that error and eventually reach a legal (or consistent) state, and resume its normal execution. A protocol that possesses the latter feature of recovering and continuing its execution starting from a legal state is also called a self-stabilizing protocol. Our recovery procedure does not require the application of an intrusive checkpointing procedure. The stable storage requirement for each process is less than that required for other proposed recovery procedures. The recovery procedure provides us with a legal protocol state, which is the global state before reaching any illegal state and before the effects of the error make other states illegal. Only a minimal number of processes affected by error propagation are required to rollback. Our recovery procedure can be used to recover from any number of transient errors in the system. Our recovery procedure has also been modeled in PROMELA, a language to describe validation models, which shows the syntactic correctness of our recovery protocol design. Finally, our procedure is compared with the existing approaches of handing the errors, and an illustrative example is provided.