International audience

The emergence of Software-Defined Networking (SDN) has marked a profound shift in the landscape of network management, revolutionizing how networks are designed, operated, and controlled. However, the task of attaining optimal routing in SDN while enhancing efficient network performance and ensuring the highest Quality of Service (QoS) still remains a challenging problem. This paper introduces a novel Distributed Teacher-Student (DTS) method that jointly harnesses the power of Multi-Agent Deep Reinforcement Learning (MADRL) with Knowledge distillation for effective QoS routing. The core idea is to transfer knowledge from a domain expert (the teacher) to an intelligent agent (the student) through a distributed learning process using DRL. This collaboration involves multiple teacher and student entities working together to decide and create optimal routes while considering the specific QoS requirements for each type of data flow. The experimental findings vividly demonstrate the effectiveness of the proposed DTS framework, revealing significant enhancements in network performance, including delay and throughput ratio. Compared to the conventional MADRL approach without the Teacher-Student framework, our DTS solution showcases a remarkable improvement of 30% in normal traffic scenarios and over 52% in situations with link failures.

In resource-constrained Internet of Things (IoT) environments like Low-power and Lossy Networks (LLNs), efficient communication protocols are essential. In this context, IP multicast protocols play a crucial role, facilitating the transmission of data packets from a single source to multiple recipients, thereby conserving bandwidth, power, and time for numerous LLN applications, such as over-the-air programming, information dissemination, and device configuration. Despite their usefulness, existing multicast solutions face several challenges, including scalability, energy efficiency, and reliability. To tackle such issues, this paper introduces Protocol Independent Multicast for LLNs (PIM-LLN). PIM-LLN employs a multicast distribution tree anchored at the border router, a multi-path data dissemination mechanism, and an efficient retransmission technique to route streams exclusively to regions with group members reducing energy consumption and bandwidth usage while improving response times and reliability. Through comprehensive simulations and public testbed experiments, we meticulously assess PIM-LLN’s performance, benchmarking it against state-of-the-art solutions under different scenarios. Our findings underscore the scalability, reliability, reduced latency, and efficient resource utilization of PIM-LLN in terms of memory, bandwidth, and energy. Notably, PIM-LLN, as compared to state-of-the-art solutions, achieves a similar level of reliability while reducing overhead by up to 50%.

International audience

Efficient and responsive traffic engineering is crucial for maintaining the robustness and reliability of Wide Area Networks (WANs). Traditional traffic engineering approaches often struggle to adapt to the dynamic and complex demands of today's network environments. To address these challenges, this paper enhances the Traffic Engineering algorithms by integrating an attention mechanism within the Edge-Path Embedding component. This significantly improves the model's adaptability and decision-making accuracy. Our comprehensive experimental evaluations demonstrate substantial improvements in terms of satisfied traffic demands and computational efficiency, highlightina the effectiveness of our approach.

Over the past few years, notable advancements have been made through the adoption of self-attention mechanisms and perceptual optimization, which have proven to be successful techniques in enhancing the overall quality of image reconstruction.Self-attention mechanisms in Vision Transformers have been widely used in neural networks to capture long-range dependencies in image data, while perceptual optimization has been shown to enhance the perceptual quality of reconstructed images.In this paper, we present a novel approach to image reconstruction by bridging the capabilities of Vision Transformer and Perceptual Compressive Sensing Networks.Specifically, we use a self-attention mechanism to capture the global context of the image and guide the sampling process, while optimizing the perceptual quality of the sampled image using a pretrained perceptual loss function.Our experiments demonstrate that our proposed approach outperforms existing state-of-the-art methods in terms of reconstruction quality and achieves visually pleasing results.Overall, our work contributes to the development of efficient and effective techniques for image sampling and reconstruction, which have potential applications in a wide range of domains, including medical imaging and video processing.

In this paper, we introduce a novel pilot-based channel estimation approach for Orthogonal Time Frequency Space (OTFS) modulation systems. Our proposed method, which involves inserting a row of pilot symbols in the zero padding region of the ZP-OTFS system, offers significant advantages such as reduced overhead and lower Peak-to-Average Power Ratio (PAPR). Notably, by spreading pilot symbols across the Doppler axis, our method provides robust and accurate channel estimation even in high mobility scenarios, a characteristic feature of environments like the TransPod system for 1000 km/h+ transportation. This is crucial, as the Doppler shifts in such scenarios could otherwise lead to substantial degradation in system performance. We investigate the channel transfer function of the TransPod guideway using the ray-tracing propagation model to illustrate this advantage. Through simulation results, we demonstrate that our proposed method outperforms the conventional pilot-based estimation method, highlighting its potential to significantly enhance OTFS system performance in practical, high-mobility scenarios. This research offers valuable insight into the potential improvements in OTFS channel estimation techniques and their real-world applications.

Due to the constant influx of multiple security attacks into the next generation of mobile communication technologies, the Third Generation Partnership Project (3GPP) has established authentication and key agreement protocol, 5-GAKA, to securely access the 5G communication services while maintaining the integrity of the underlying network. However, some recent findings pointed out that 5G-AKA has many drawbacks, including perfect forward secrecy violations, malicious Serving Network (SN) attacks, desynchronization attacks, privacy theft, stolen device, and denial of service (DoS) attacks when the user uses roaming mobile services. Considering the drawbacks of current 5G communication protocols and the necessity to facilitate additional security, a provably secure and lightweight protocol for 5G communication (PSLP-5G) is introduced. The PSLP-5G's security is guaranteed using the Scyther tool and Real-Or-Random (ROR) logic. Furthermore, performance comparisons are made to show how much lighter the PSLP-5G is than its counterparts. Additionally, the PSLP-5G's suitability for use in real-time applications is demonstrated by comparing the network performance of PSLP-5G and its counterparts using the Network Simulator tool NS3.

Artificial intelligence (AI) has transformed the field of natural language processing, enabling substantial advances in understanding, interpreting, and producing human language. The ability of AI to find new solutions to difficult linguistic expressions has led to the birth of sophisticated language models such as ChatGPT. This model uses cutting-edge deep learning algorithms to produce high-quality, human-like writing in response to natural language inputs. ChatGPT has an amazing capacity to recognize context, evaluate sentiment, and provide coherent and appropriate replies. It has a wide range of applications, from virtual assistants and customer support bots to language translation and content development. Therefore, understanding its backend has become essential. In this paper, we summarize the key principles underlying the operation of ChatGPT's back-end. This study is required reading for ChatGPT researchers because it covers critical aspects of the ChatGPT backend. It includes essential information for researchers looking to improve ChatGPT's performance or create new language models based on its architecture.

The progress achieved in transportation systems and artificial intelligence has amplified the use of intelligent transportation systems and Autonomous Vehicles (AVs). Indeed, AV systems have attracted much research in recent years, which enabled multiple autonomous driving tasks, including scene understanding, visual prediction, decision-making, and communication. The latter may create a bottleneck in low-resource autonomous driving systems that send the collected images to remote edge servers for processing and decision-making. Such an issue can be addressed by compressing the images in the AV and ensuring a good-quality reconstruction at the edge. In this paper, we propose a deep neural network for Compressed Sensing (CS) based image reconstruction that integrates image semantic perception to improve the reconstruction process for visual prediction tasks. The reconstruction process is optimized using a perception-inspired loss in an end-to-end model learning process. The trained model is evaluated on autonomous driving car datasets. Obtained experimental results outperform state-of-the-art approaches in terms of both image reconstruction quality and processing time. Finally, we perform semantic urban scene segmentation on the reconstructed image to evaluate reconstruction quality for visual task prediction. Obtained results on three semantic urban scene datasets demonstrate the efficiency of the proposed approach.

The optimal transmit power that maximizes energy efficiency (EE) in Longe Range (LoRa) networks is investigated by using the deep learning (DL) approach. Particularly, the proposed artificial neural network (ANN) is trained two times; in the first phase, the ANN is trained by the model-based data which are generated from the simplified system model while in the second phase, the pre-trained ANN is re-trained by the practical data. Numerical results show that the proposed approach outperforms the conventional one which directly trains with the practical data. Moreover, the performance of the proposed ANN under both partial and full optimum architecture are studied. The results depict that the gap between these architectures is negligible. Finally, our findings also illustrate that instead of fully re-trained the ANN in the second training phase, freezing some layers is also feasible since it does not significantly decrease the performance of the ANN.

The system-level performance of multi-gateway downlink long-range (LoRa) networks is investigated in the present paper. Specifically, we first compute the active probability of a channel and the selection probability of an active end-device (ED) in the closed-form expressions. We then derive the coverage probability (Pcov) and the area spectral efficiency (ASE) under the impact of the capture effects and different spreading factor (SF) allocation schemes. Our findings show that both the Pcov and the ASE of the considered networks can be enhanced significantly by increasing both the duty cycle and the transmit power. Finally, Monte-Carlo simulations are provided to verify the accuracy of the proposed mathematical frameworks.

In the present paper a closed-form framework is derived for the analysis and optimization of the coverage probability (Pcov) and of the area spectral efficiency (ASE) in long-range (LoRa) networks. The proposed framework exploits stochastic geometry tools to associate the Pcov and the ASE to the end device (ED) transmit power and to the ED density. The analysis reveals the trends of the Pcov and of the ASE curves, with respect to both of the two parameters, while the robustness of the framework holds even at the asymptotic cases. Building upon the derived framework, the analysis demonstrates that no joint global optimum exists that jointly maximizes the Pcov over both parameters, suggesting that the optimization of the Pcov must be performed separately, for the two key network parameters considered. As opposed to that, the analysis demonstrates that a set of global optima exists that jointly maximize the ASE over both parameters, and these global maxima are subsequently derived in closed form. Thus, the derived framework fully characterizes the performance of LoRa networks, while defining in closed form the optimal points of operation that can be proven of significant value, for the transceiver and network design, of practical LoRa networks.

Fifth generation and beyond networks are envisioned to support multi industrial Internet of Things (IIoT) applications with a diverse quality-of-service (QoS) requirements. Network slicing is recognized as a flagship technology that enables IIoT networks with multiservices and resource requirements by allowing the network-as-infrastructure transition to the network-as-service. Motivated by the increasing IIoT computational capacity, and taking into consideration the QoS satisfaction and private data sharing challenges, federated reinforcement learning (RL) has become a promising approach that distributes data acquisition and computation tasks over distributed network agents, exploiting local computation capacities and agent's self-learning experiences. This article proposes a novel deep RL scheme to provide a federated and dynamic network management and resource allocation for differentiated QoS services in future IIoT networks. This involves IIoT slices resource allocation in terms of transmission power (TP) and spreading factor (SF) according to the slices QoS requirements. Toward this goal, the proposed deep federated Q-learning (DFQL) is reached into two main steps. First, we propose a multiagent deep Q-learning-based dynamic slices TP and SF adjustment process that aims at maximizing self-QoS requirements in term of throughput and delay. Second, the deep federated learning is proposed to learn multiagent self-model and enable them to find an optimal action decision on the TP and the SF that satisfy IIoT virtual network slice QoS reward, exploiting the shared experiences between agents. Simulation results show that the proposed DFQL framework achieves efficient performance compared to the traditional approaches.

In this work, coverage probability (Pcov) and area spectral efficiency (ASE) of LoRa networks with multiple classes of quality-of-service (QoS) of end-devices (EDs) are investigated. In particular, an approximated but tractable mathematical framework is proposed to compute a recent definition of the Pcov. Based on the proposed framework, the closed-form expressions of both Pcov and ASE are provided. Moreover, the trends of both metrics are unveiled that respect to some key parameters such as the density of the EDs, the transmit power, and the spreading factor (SF). Our findings show that depending on networks parameters, the ASE is either a unimodal function or monotonically increasing as a function of EDs' density. In addition, the optimal value of EDs' density that maximizes the ASE are computed in closed-form expression too. Finally, Monte Carlo simulations are provided to verify the correctness of our framework.

The future fifth-generation (5G) networks are expected to support a huge number of connected devices with various and multitude services having different quality of service (QoS) requirements. Communication in Industry 4.0 is one of the flagships and special applications of the 5G due to the specificity of the industrial environment as well as the variety of its services such as safety communication, robot's communications, and machine monitoring. In this context, we propose a new resource allocation for the future Industry 4.0 based on software-defined networking and network function virtualization technologies, machine learning tools and the slicing paradigm where each slice of the network is dedicated to a category of services having similar QoS requirement level. In this article, the proposed solution ensures the allocation of the resources to the slices depending on their requirements in terms of bandwidth, delay, and reliability. Toward this goal, our solution is performed in three main steps: first, Internet of Things (IoT) devices assignment to the slices step based on online Gaussian mixture model clustering algorithm, second, inter-slices resources reservations step based on mini-batch gradient descent, and third, intra-slices resources allocations based on the max-utility algorithm. We have performed extensive simulations in a realistic industrial scenario using NS3 simulator. Numerical results show the effectiveness of our proposed solution in terms of reducing packet error rate, energy consumption, and in terms of increasing the percentage of served devices in delay comparing to the traditional approaches.

The massive growth of the Internet of Things (IoT) poses important challenges on providing IoT devices with specific quality of service (QoS) requirements in terms of urgency and reliability over long distances and Long Range Wide Area Network (LoRaWAN). Due to the diversity of these services and the increasing complexity in large scale IoT networks, software defined networking (SDN) alongside network slicing are needed to increase flexibility in managing network slices and providing IoT networks with an optimized parameters configuration. Therefore, we propose in this paper a SDN-based network slicing architecture for LoRaWAN where network slices are virtually deployed and isolated over LoRa physical gateways. Moreover, we aim to improve large scale network configuration by proposing TOPG, a slice-based optimization that improves LoRa parameters configuration based on QoS thresholds of each slice. Simulation results performed over NS3, highlight the utility of the proposed optimization in improving the network performance of LoRa slices in terms of reliability and respecting QoS thresholds in IoT dense deployments.

The massive growth of the Internet of Things (IoT) poses important challenges on network operators to support billions of IoT devices connected through the cloud with each having constrained battery life and computational capacity. To support these requirements over long distances, Long Range Wide Area Network (LoRaWAN), is now widely being deployed with the promise to support an all-connected world with numerous IoT applications. In large scale access networks, supporting urgent and reliable communications with their QoS demands becomes more challenging. Hence, network slicing within an SDN-based architecture brings numerous advantages to solve this problem by easily managing network resources and reserving part of the latter for urgent traffic and avoiding its performance degradation due to congestion. In this paper, we tackle the raised questions regarding scalability limitations by proposing a distributed slicing strategy based on coalitional game and matching theory over an SDN-based LoRaWAN architecture. In this context, resource reservation for LoRa slices and configuration optimization are performed closer to the edge at the gateway level. Simulation results performed over NS3 highlight the utility of the distributed slicing strategy in respecting quality of service (QoS) thresholds in terms of delay, throughput, energy consumption and improving reliability while providing complete isolation between LoRa slices.

Network function virtualization (NFV) will simplify deployment and management of network and telecommunication services. NFV provides flexibility by virtualizing the network functions and moving them to a virtualization platform. In order to achieve its full potential, NFV is being extended to mobile or wireless networks by considering virtualization of radio functions. A typical network service setup requires the allocation of a virtual network function-forwarding graph (VNF-FG). A VNF-FG is allocated considering the resource constraints of the lower infrastructure. This topic has been well-studied in existing literature, however, the effects of variations of networks over time have not been addressed yet. In this paper, we provide a model of the adaptive and dynamic VNF allocation problem considering also VNF migration. Then we formulate the optimization problem as an integer linear programming (ILP) and provide a heuristic algorithm for allocating multiple VNF-FGs. The idea is that VNF-FGs can be reallocated dynamically to obtain the optimal solution over time. First, a centralized optimization approach is proposed to cope with the ILP-resource allocation problem. Next, a decentralized optimization approach is proposed to deal with cooperative multi-operator scenarios. We adopt AD3, an alternating direction method of multipliers-based algorithm, to solve this problem in a distributed way. The results confirm that the proposed algorithms are able to optimize the network utilization, while limiting the number of reallocations of VNFs which could interrupt network services.

Network Function Virtualization (NFV) is emerging as an efficient mean to deploy and manage network and telecommunication services. With wireless access networks, NFV has to take into account the radio resources at wireless nodes in order to provide an end-to-end optimal virtual network function (VNF) allocation. This topic has been well-studied in existing literature, however, the effects of variations of networks over time have not been addressed yet. In this paper, we provide a model of the adaptive and dynamic VNF allocation problem considering VNF migration. Moreover, we also consider service function chains (SFCs) with QoS constraints. Then we formulate the optimisation problem as an Integer Linear Programming (ILP) and provide a heuristic algorithm for allocating multiple SFCs. The proposed approach allows SFCs to be reallocated so as to obtain the optimal solution over time. The results confirm that the proposed algorithm is able to optimize the network utilization while limiting the reallocation of VNFs which could interrupt services.

Network Function Virtualization (NFV) is deemed as a mean to simplify deployment and management of network and telecommunication services. With wireless access networks, NFV has to take into account the radio resources at wireless nodes in order to provide an end-to-end optimal virtual network function (VNF) allocation. This topic has been well-studied in existing literature, however, the effects of variations of networks over time have not been addressed yet. In this paper, we provide a model of the adaptive and dynamic VNF allocation problem considering VNF migration. Then we formulate the optimisation problem as an Integer Linear Programming (ILP) and provide a heuristic algorithm for allocating multiple service function chains (SFCs). The proposed approach allows SFCs to be reallocated so as to obtain the optimal solution over time. The results confirm that the proposed algorithm is able to optimize the network utilization while limiting the reallocation of VNFs which could interrupt services.

A fundamental problem in designing sensors network is locating their position. The data collected from the sensors can be used to detect, track and organize objects of interest. In this paper, we present and evaluate an improvement of the famous DV-HOP algorithm in order to increase the localization accuracy and reduce energy consumption. The benefits of the suggested algorithm are twofold. First, it uses a new technique for solving an N-equation system and a weighted least squares method (WLS) to minimize the error of the expected distance between anchor and unknown nodes. Second, this method uses the hop-size average of the anchor node, which is computed by unknown nodes, to reduce the overall communication cost between nodes. This yields a significant reduction in both energy consumption and execution time. The performance of our proposed approach was evaluated and compared to other classical algorithms. Results show that significant enhancement is achieved within the proposed algorithm when measuring different metrics such as energy, execution time and localization error while varying simulation parameters such as the total number of nodes, percentage of anchor node and communication range.

Small-cells are rapidly emerging as the mobile operators' choice to provide additional capacity in current and future mobile networks. However, in order to fully deliver on their promises, small-cells need to address severe interference control and coordination challenges. By centralizing base-band processing in large high-volume computing infrastructures, Cloud-RAN can effectively enable advanced coordination features for dense small-cells deployments. Unfortunately, Cloud-RAN tight bandwidth and latency requirements have made optical fiber the most common solution for the links interconnecting remote radio heads (RRHs) with the base band units (BBUs), i.e. the fronthaul. Recent advances in microwave communications are making wireless fronthauls a viable option especially in dense urban environments where fiber fronthauls could be too rigid for accommodating highly dynamic traffic patterns. In this paper, we provide a novel formulation for the BBU Placement problem where BBU pools are placed at the edges of the network, possibly co-located with macro-cells, and a reconfigurable wireless fronthaul is used in order to provide RRHs with connectivity. To the best of our knowledge this is the first work to tackle the BBU placement problem over a reconfigurable substrate network with mmWave links. We also propose a BBU Placement heuristics, and we evaluate it using a numerical simulator.

The Internet of Things (IoT) represents the next wave in networking and communication which will bring by 2020 tens of billions of Machine-to-Machine (M2M) devices connected through the internet. Hence, this rapid increase in Machine Type Communications (MTC) poses a challenge on cellular operators to support M2M communications without hindering the existing Quality of Service for already established Human-to-Human (H2H) communications. LTE-M is one of the candidates to support M2M communications in Long Term Evolution (LTE) cellular networks. In this paper, we appraise and present an in depth performance evaluation of LTE-M based on cross-layer network metrics. Compared with LTE Category 0 previously released by 3GPP for MTC, simulation results show that LTE-M offers additional advantages to meet M2M communication needs in terms of wider coverage, lower throughput, and a larger number of machines connected through LTE network. However, we show that LTE-M is not yet up to the level to meet future applications requirements regarding a near-zero latency and an advanced Quality of Service (QoS) for this massive number of connected Machine Type devices (MTDs).

Network function virtualization (NFV) sits firmly on the networking evolutionary path. By migrating network functions from dedicated devices to general purpose computing platforms, NFV can help reduce the cost to deploy and operate large IT infrastructures. In particular, NFV is expected to play a pivotal role in mobile networks where significant cost reductions can be obtained by dynamically deploying and scaling virtual network functions (VNFs) in the core network. However, in order to achieve its full potential, NFV needs to extend its reach also to the radio access segment. Here, mobile virtual network operators shall be allowed to request radio access VNFs with custom resource allocation solutions. Such a requirement raises several challenges in terms of performance isolation and resource provisioning. In this work, we formalize the wireless VNF placement problem in the radio access network as an integer linear programming problem and we propose a VNF placement heuristic, named wireless network embedding (WiNE), to solve the problem. Moreover, we present a proof-of-concept implementation of an NFV management and orchestration framework for enterprise WLANs. The proposed architecture builds on a programmable network fabric where pure forwarding nodes are mixed with radio and packet processing capable nodes.

Network Function Virtualization (NFV) is emerging as one of the most innovative concepts in the networking landscape. By migrating network functions from dedicated mid-dleboxes to general purpose computing platforms, NFV can effectively reduce the cost to deploy and to operate large networks. However, in order to achieve its full potential, NFV needs to encompass also the radio access network allowing Mobile Virtual Network Operators to deploy custom resource allocation solutions within their virtual radio nodes. Such requirement raises several challenges in terms of performance isolation and resource provisioning. In this work we formalize the Virtual Network Function (VNF) placement problem for radio access networks as an integer linear programming problem and we propose a VNF placement heuristic. Moreover, we also present a proof-of-concept implementation of an NFV management and orchestration framework for Enterprise WLANs. The proposed architecture builds upon a programmable network fabric where pure forwarding nodes are mixed with radio and packet processing nodes leveraging on general computing platforms.

Streaming over multimedia WSN (MWSN) in urban environment is challenging due to many issues among which spectrum scarcity and high radio interference. Such conditions make it difficult to ensure high bandwidth, low transmission delay and low packet losses required for real time multimedia streaming applications. In this paper, we propose COMUS a COgnitive radio solution for MUltimedia streaming over wireless Sensor networks which uses both cognitive radio technology and clustering mechanism to enhance spectrum and energy efficiency. In COMUS we consider clustering the MWSN nodes into different clusters to ensure low energy consumption. Furthermore, based on the nodes geographical position and the actual and the forecasted channel availability, we aim to ensure stable clusters forming. The multimedia streaming from a particular source node to the sink node, require a physical channel selection to perform the corresponding routing task. Thus, in COMUS we propose an efficient channel selection to prevent frequent channel switching which considers the PU (Primary User) activity forecasts. Our simulation results show that COMUS outperforms the two existing pioneering mechanisms called SEARCH and SCEEM and this in terms of providing higher video quality (PSNR and frame rate), lower end-to-end transmission delay and lower frame loss ratio under varied spectrum conditions.

No abstract available.

Nowadays, the channel selection is a challenging task in cognitive radio because of the high radio activity and the preemptive priority of the licensed user, called primary user (PU). In this paper, we propose a DIStributed channel Selection mechanism for efficient content dissemination in COgnitive RaDio ad-hoc networks (DISCORD). DISCORD selects the most appropriate channel for content dissemination based on the PU channel occupancy and the importance of cognitive radio neighbors in the network. Indeed, a sender peer in DISCORD forecasts the channel primary user activity by the mean of the actual and statistical estimation of the channels occupancy and selects the most stable one. Moreover, it makes use of a new Social Networks Analysis (SNA) inspired metric to select the appropriate neighbors for high content dissemination in the network. The simulation results using NS2 shows that Discord presents highest performance in terms of interferences, PU priority respect and content delivery ratio in multi-hop CRNs comparing to four related protocols.

Video content delivery for vehicular ad hoc networks (VANET) under dense network conditions poses non trivial issues because of the scarcity and volatility of the wireless medium. In this context, video communication is envisioned to be of high benefit for traffic management as well as for providing value-added entertainment and advertising services. In this paper, we propose a new mechanism for efficient video streaming over VANET. The proposed mechanism selects a minimum sub-set of rebroadcaster vehicles in order to reduce interferences and achieve high video quality. The vehicles are ranked based on their strategic location in the network and their capacity to reach other vehicles using a new centrality metric inspired from the Social Network Analysis (SNA), called dissemination capacity. Through simulations, we compared our mechanism with the multichannel vehicular communication standard IEEE 1609.4 and another pioneering video streaming mechanism over VANET. The performance evaluation shows that it outperforms the abovementioned mechanisms by providing higher video delivery ratio, lower end- to-end transmission delay and lower frame loss ratio in both fully and intermittently connected networks.

During the last decade, we witnessed a rapid growth in deployment of pull-based P2P streaming applications. In these applications, each node selects some other nodes as its neighbors and requests streaming data from them. This scheme allows eliminating data redundancy and recovering from data loss, but it pushes the complexity to the receiver node side. In this paper, we theoretically study the scheduling problem in pull-based P2P video streaming and we model it as an assignment problem. Then, we propose AsSched, new scheduling algorithm for layered streaming, in order to optimize the throughput and the delivery ratio of the system. In second time, we derive an optimal algorithm (NAsSched) for non layered streaming. The results of simulations show that our algorithms significantly outperform classic scheduling strategies especially in stern bandwidth constraints.

Although IP multicast techniques were proposed a long time ago and despite of their advantages, they are still not widely deployed due to the absence of multicast support in some routers/domains and inter-domain management issues. On the other hand, in the most of recent internet applications, where the average consumed bandwidth is measured by hundreds of Kbits per second and where the support of large-scale distribution is important, the IP multicast becomes more than a necessity. In this paper, we propose a new approach for extending the scope of IP multicast in overlay applications. We selected some overlay nodes to be used as fan-out multicast nodes and then created an IP multicast islands around each fan-out node. These multicast islands are connected with each other using unicast overlay links. This selection of fan-out nodes is based on a distributed version of K-means algorithm and GNP (Global Network Positioning), a distributed technique to measure the distance between nodes. We further propose a preventive fault tolerance mechanism for packet loss across islands. Finally, the simulation results verify the optimality of our approach in terms of link stress minimization and end-to-end delay reduction.