Renewable energy has become a viable alternative to fossil fuels owing to its environmental benefits. However, its inherent uncertainty pose significant challenges. Demand response mechanisms have been developed to address these issues, facilitating renewable energy integration through consumer-side flexible resources. However, these mechanisms often affect consumer satisfaction, necessitating precise measurement and control of these impacts. In this paper, we propose a two-stage electricity trading and load dispatch optimization model aimed at reducing carbon emission by promoting renewable energy accommodation, and the proposed optimization model takes into account multi-category energy consumer satisfaction. We begin by classifying consumers into distinct categories and designing tailored satisfaction functions that reflect their unique power consumption preferences. The electricity trading and load dispatch processes are formulated as a two-stage optimization problem, which is then transformed into Markov decision processes. A model-free framework applying two state-of-the-art deep reinforcement learning algorithms is proposed to solve the optimization problem without requiring complex environmental modeling and prior knowledge. Numerical results demonstrate that the proposed framework outperforms benchmark algorithms regarding both consumer satisfaction preservation and carbon emission reduction.

The rapidly increasing popularity of immersive multimedia services such as live holographic communication represents the future trend of extended reality (XR) applications. However, the realization of such immersive and interactive experiences is limited by the lack of fundamental understanding of how different user behaviours and environmental factors jointly affect the overall quality of experience (QoE). In particular, compared with the media adaptation mechanisms applied in conventional video applications, considerably more independent factors may influence user QoE in these applications, including both human- and network-related factors. In this paper, we investigate the fundamental design principles of dynamic media adaptation methods for live holographic communication by holistically considering these factors. Specifically, a machine learning-based scheme is introduced to facilitate intelligent adaptation of both the frame quality (resolution level) and frame rate according to specific contexts, such as the user intent/behaviour (including object motion patterns and user movements) and real-time network conditions. Extensive real-world experiments are conducted to assess the feasibility and performance of the proposed method, and comparisons with state-of-the-art methods are performed. The results indicate that the proposed approach can effectively satisfy user intent, with increased user QoE.

Driven by the rapid advancement of XR technologies, an increasing demand has emerged for enabling interactive holographic real-time applications of multiple users. However, in wireless systems, the challenges of practically delivering a satisfactory experience for such applications arise not only from the high bandwidth and computational demands of holographic content but also from the need to effectively integrate and transmit additional sensory data, such as haptic feedback, to convey accurate and human-perceivable semantics. In this paper, we propose an architecture that leverages the computational capabilities of 5G edge computing to collaborate with multi-user haptic-integrated holographic video applications. We detail the components of this solution architecture, including haptic-integrated holographic video capture and streaming, and a simple reinforcement learning algorithm to synchronise multiuser frames at the MEC server. A prototype of a representative hand-touch social interaction application was designed, implemented, and measured to exemplify the potential of human-perceivable, haptic-integrated holographic video applications.

The performance of collective communication schedules is crucial for the efficiency of machine learning jobs and GPU cluster utilization. Existing open-source collective communication libraries (such as NCCL and RCCL) rely on fixed schedules and cannot adjust to varying topology and model requirements. State-of-the-art collective schedule synthesizers (such as TECCL and TACCL) utilize Mixed Integer Linear Program for modeling but encounter search space explosion and scalability challenges. In this paper, we propose SyCCL, a scalable collective schedule synthesizer that aims to synthesize near-optimal schedules in tens of minutes for production-scale machine-learning jobs. SyCCL leverages collective and topology symmetries to decompose the original collective communication demand into smaller sub-demands within smaller topology subsets. SyCCL proposes efficient search strategies to quickly explore potential sub-demands, synthesizes corresponding sub-schedules, and integrates these sub-schedules into complete schedules. Our 32-A100 testbed and production-scale simulation experiments show that SyCCL improves collective performance by up to 127% while reducing synthesis time by 2 to 4 orders of magnitude compared to state-of-the-art efforts.

Supporting ubiquitous deployment of built-in Internet service with Software Defined Networking (SDN), Network Function Virtualization (NFV), and Low Earth Orbit (LEO) satellite constellations has been widely accepted as one of the key technologies for the next-generation communication services. By integrating terrestrial and space network capabilities, new design features are introduced to existing network ecosystems. For instance, terrestrial Virtual Network Functions (VNFs) can now be hosted on satellites, utilizing satellite highways. This requires expanding the roles of the control units, originally responsible for terrestrial data planes, to include space-based counterparts. As a result, seamless integration of Service Function Chains (SFCs) across satellite constellations and terrestrial control units becomes a challenge due to the topology dynamics caused by high-speed LEO satellites. In this paper, we propose the Geosynchronous Service Function Chaining (GSFC) scheme to facilitate programmable, Internet SFC operations based on LEO satellite network environments. The key idea is to cluster adjacent LEO satellites to represent logical VNF containers at the fixed positions, where the initial VNFs at the region are continuously filled up by the traversing satellite payload functions in a predictable manner. With this design, the ground-based controllers can maintain the space-terrestrial SFCs without being affected by the constantly shifting satellite VNFs, and thereby large-scale and complex recalculation for the routing policies is avoided. The design principle introduces a ground-breaking approach to space Internet protocol stacks, facilitating robust routing for SFC operations across integrated space and terrestrial networks. Our simulation results verify the feasibility of the proposed GSFC-based VNF orchestration mechanism and reveal the trade-offs in both data and control plane performance.

As a representative application paradigm for supporting volumetric video, point cloud is able to capture three-dimensional spatial information of objects, offering dynamic and immersive experiences to end users. Compared to the conventional bit-to-bit communication principle, the design of semantic communication ingeniously utilizes joint source and channel coding to transmit semantic features instead. By taking into account the common content delivery requirements of low bandwidth consumptions and low latency, while maintaining high resolutions, we present in this paper a novel framework called Spatial-Temporal Semantic Point Cloud Transmission (ST-SPCT). Compared to the existing point cloud compression and semantic communication methods that extract and reconstruct features only in the spatial dimension, the ST-SPCT simultaneously extracts and reconstructs spatial-temporal semantic features more deeply, thus significantly reducing computational time and data volume, while ensuring negligible compromises in peak signal-to-noise ratio (PSNR), chamfer distance (CD) metrics, and frame continuity according to our systematic experiment results.

Volumetric video streaming, an innovative media application, facilitates the real-time and immersive teleportation of individuals or objects into the virtual environment of the audience. Unlike conventional video streaming applications, volumetric content is particularly vulnerable to network fluctuations, which can lead to performance degradation such as reduced FPS, delayed frame delivery. In this research, we introduce an eBPF-based network function that duplicates packets along the pathways between network nodes, ensuring timely packet delivery amid network instability. Furthermore, we propose a path elimination algorithm to discard paths incapable of delivering frames within the target latency. Our implementation and evaluation validate the rapid and robust performance achieved across various resolution levels.

Efficient resource management and orchestration are essential to maximise the benefits of network slicing in 5G and beyond networks. This paper presents NFScaler, an Artificial Intelligence (AI)-powered 5G network function scaler that can perform a zero-shot sim2real transfer. The proposed NFScaler automatically adjusts the number of 5G network function instances running in a network slice based on changes in user traffic to meet the QoS requirement of the slice while reducing the resource consumption (i.e. CPU and power). The NFScaler consists of two main parts: a domain randomiser and a deepreinforcement learning (DRL) agent. The domain randomiser randomises the dynamic parameters of a simulation to provide the DRL agent with a range of simulation environments. The DRL agent interacts with both the original simulation and the randomised simulation to learn a generalisable autoscaling policy. The performance of the proposed NFScaler is evaluated in a live 5G network and compared with the threshold-based KEDA scaler, which is the industry standard event-based autoscaler used in cloud native environments, as well as a DRL method that does not involve sim2real transfer. The experimental findings demonstrate that NFScaler effectively guarantees the QoS requirements of the network slice, outperforming the benchmark methods. In particular, NFScaler achieves an improvement of almost 40% in QoS (throughput) performance compared to KEDA Scaler. Furthermore, the proposed NFScaler allocates a significantly lower number of CPU cores to the user plane of the network slice, with an average of 75% fewer cores than when the user plane is hosted on a dedicated bare metal server. In addition, the NFScaler reduces the power consumption of the user plane of the network slice by an average of 45% compared to hosting the user plane on a dedicated bare metal server.

Open Radio Access Networks (O-RANs) have revolutionized the telecom ecosystem by bringing intelligence into disaggregated RAN and implementing functionalities as Virtual Network Functions (VNF) through open interfaces. However, dynamic traffic conditions in real-life O-RAN environments may require necessary VNF reconfigurations during run-time, which introduce additional overhead costs and traffic instability. To address this challenge, we propose a multi-objective optimization problem that minimizes VNF computational costs and overhead of periodical reconfigurations simultaneously. Our solution uses constrained combinatorial optimization with deep reinforcement learning, where an agent minimizes a penalized cost function calculated by the proposed optimization problem. The evaluation of our proposed solution demonstrates significant enhancements, achieving up to 76% reduction in VNF reconfiguration overhead, with only a slight increase of up to 23% in computational costs. In addition, when compared to the most robust O-RAN system that doesn’t require VNF reconfigurations, which is Centralized RAN (C-RAN), our solution offers up to 76% savings in bandwidth while showing up to 27% overprovisioning of CPU.

Integrating Low Earth Orbit (LEO) satellites with terrestrial network infrastructures to support ubiquitous Internet service coverage has recently received increasing research momentum. One fundamental challenge is the frequent topology change caused by the constellation behaviour of LEO satellites. In the context of Software Defined Networking (SDN), the controller function that is originally required to control the conventional data plane fulfilled by terrestrial SDN switches will need to expand its responsibility to cover their counterparts in the space, namely LEO satellites that are used for data forwarding. As such, seamless integration of the fixed control plane on the ground and the mobile data plane fulfilled by constellation LEO satellites will become a distinct challenge. For the very first time in the literature, we propose in this paper the Virtual Data-Plane Addressing (VDPA) scheme by leveraging IP addresses to represent virtual switches at the fixed space locations which are periodically instantiated by the nested LEO satellites traversing them in a predictable manner. With such a scheme the changing data-plane network topology incurred by LEO satellite constellations can be made completely agnostic to the control plane on the ground, thus enabling a native approach to supporting seamless communication between the two planes. Our simulation results prove the superiority of the proposed VDPA based flow rule manipulation mechanism in terms of control plane performance.

Service Function Chaining (SFC) is an emerging network technology based on network function virtualisation (NFV), which facilitates in-network data processing during traffic steering. SFC optimisations can be broadly categorised into centralised and distributed solutions, each with advantages and disadvantages. In this paper, we propose a flexible SFC optimisation scheme based on network cluster partitioning, where cluster-based information sharing and decision-making are applied for building optimised SFCs in real-time. Such a scheme can be flexibly adapted based on cluster numbers and sizes. The key challenge is how necessary intra-cluster information is shared across clusters for performance optimisation while retaining privacy and low signalling complexity. According to our simulation experiments based on real network topologies, our proposed solution outperforms the state-of-the-art benchmark by about 20%.

BGP route hijacking is a critical threat to the Internet. Existing works on path hijacking detection firstly monitor the routes of the whole network and then directly trigger a suspicious alarm if the link has not been seen before. However, these naive approaches will cause false positive identification and introduce unnecessary verification overhead. In this work, we propose Metis, a matching-and-prediction system to filter out normal unseen links. We first use a matching method with three rules to find out suspicious links if there is an unseen AS. Otherwise, we propose using a neural network to make a prediction based on the AS information at each end of the link and further quantify the suspicion level. Our large-scale simulation results show that Metis can achieve precision and recall of over 80% for detecting fake AS-PATHs. Moreover, our deployment experiences show that compared to state-of-the-art system, Metis can save 80% overhead.

Smart grid integrates the distributed energy resources such as renewable energy with massive information to facilitate the flow of energy in the industries. The renewable energy accommodation is one of the key issues to achieve the energy efficiency in smart grid, which is difficult to obtain dynamic optimal policies due to the intermittency of renewables. To capture statuses of renewable energy for decision-making, large amounts of information in heterogeneous forms are collected by massive end devices deployed in smart grid. Such information not only provides fruitful features for existing learning based algorithms but also incurs high computation complexity. Besides, such heterogeneous data may also contain missing values, which may result in wrong policies by existing algorithms. In this article, a novel cloud-edge-end orchestrated computing scheme is proposed to efficiently repair missing values and obtain optimal policies in two separate layers. In the first layer, deep learning based algorithms deployed can perceive the characteristics and repair the missing values. In the second layer, deep reinforcement learning based algorithms are employed to obtain optimal policies. Simulations on the real power grid dataset illustrate the effectiveness of proposed fault-tolerant renewable energy accommodation algorithm.

Live holographic teleportation is an emerging media application that allows Internet users to communicate in a fully immersive environment. One distinguishing feature of such an application is the ability to teleport multiple objects from different network locations into the receiver's field of view at the same time, mimicking the effect of group-based communications in a common physical space. In this case, live teleportation frames originated from different sources must be precisely synchronised at the receiver side to ensure user experiences with eliminated perception of motion misalignment effect. For the very first time in the literature, we quantify the motion misalignment between remote sources with different network contexts in order to justify the necessity of such frame synchronisation operations. Based on this motivation, we propose HoloSync, a novel edge-computing-based scheme capable of achieving controllable frame synchronisation performances for multi-source holographic teleportation applications. We carry out systematic experiments on a real system with the HoloSync scheme in terms of frame synchronisation performances in specific network scenarios, and their sensitivity to different control parameters.

Forwarding path selection for data traffic is one of the most fundamental operations in computer networks, whose performance drastically impacts both transmission efficiency and reliability in network domains. Although deep reinforcement learning (DRL) has attracted considerable attention for path selection instead of hand-tuned heuristics, few works have considered how to exploit graph-structured information in networks to improve routing and forwarding efficiency. In fact, generating routes is essentially a process for finding a subgraph in a graph-structured network. To this end, this paper proposes an effective and novel graph embedding-based DRL framework for adaptive path selection (termed GRL-PS), aiming at reducing end-to-end (E2E) latency and promoting network throughput while maintaining stability in dynamically changing environments. Specifically, graph representation learning (GRL) is deployed as an effective enabler for the DRL agent to learn the relational knowledge of interacting entities for route decisions in networks. However, training such an agent in a dynamically changing environment encounters a knowledge acquisition bottleneck, since the DRL agent is always forced to learn every task from scratch. To improve the adaptation of behaviors and acquire skills beyond what the source policy can teach, we introduce potential-based reward shaping as a means of knowledge transfer to guide the agent in unfamiliar conditions with sparse rewards. Experimental results show that compared with baseline methods, our solution can achieve nearly-optimal performance with both latency and throughput, especially in large-scale dynamic networks.

With the development of edge-cloud computing technologies, distributed data centers (DCs) have been extensively deployed across the global Internet. Since different users/applications have heterogeneous requirements on specific types of ICT resources in distributed DCs, how to optimize such heterogeneous resources under dynamic and even uncertain environments becomes a challenging issue. Traditional approaches are not able to provide effective solutions for multi-dimensional resource allocation that involves the balanced utilization across different resource types in distributed DC environments. This paper presents a reinforcement learning based approach for multi-dimensional resource allocation (termed as NESRL-MRM) that is able to achieve balanced utilization and availability of resources in dynamic environments. To train NESRL-MRM’s agent with sufficiently quick wall-clock time but without the loss of exploration diversity in the search space, a natural evolution strategy (NES) is employed to approximate the gradient of the reward function. To realistically evaluate the performance of NESRL-MRM, our simulation evaluations are based on real-world workload traces from Amazon EC2 and Google datacenters. Our results show that NESRL-MRM is able to achieve significant improvement over the existing approaches in balancing the utilization of multi-dimensional DC resources, which leads to substantially reduced blocking probability of future incoming workload demands.

Distributed mobility management (DMM) solution is proposed to address the downsides of centralized mobility management protocols. The standard DMM is proposed for flat architectures and always selects the anchor point from access layer. Numerical analysis is used in this paper to show that dynamic anchor point selection can improve the performance of standard DMM in terms of packet signalling and delivery cost. In next step, an SDN-based DMM solution that we refer to as SD-DMM is presented to provide dynamic anchor point selection for hierarchical mobile network architecture. In SD-DMM, the anchor point is dynamically selected for each mobile node by a virtual function implemented as an application on top of the SDN controller which has a global view of the network. The main advantages of SD-DMM is to decrease packet delivery cost.

Open radio access network (Open-RAN) is becoming a key component of cellular networks, and therefore optimizing its architecture is vital. The Open-RAN is a distributed architecture that lets the virtualized networking functions be split between Distributed Units (DU) and Centralized Units (CUs); as a result, there is a wide range of design options. We propose an optimization problem to choose the split points. The objective is to balance the load across CUs as well as midhaul links by considering delay requirements. The resulting formulation is an NP-hard problem that is solved with a novel heuristic algorithm. Performance evaluation shows that the gap between optimal and heuristic solutions does not exceed 2%. An in-depth analysis of different centralization levels shows that using multi-CUs could reduce the total bandwidth usage by up to 20%. Moreover, multipath routing can improve the result of load balancing between midhaul links while increasing bandwidth usage.

In data center networks, traffic needs to be distributed among different paths using traffic optimization strategies for mixed flows. Most of the existing strategies consider either distributed or centralized mechanisms to optimize the latency of mice flows or the throughput of elephant flows. However, low network performance and scalability issues are intrinsic limitations of both strategies. In addition, the current elephant flow detection methods are inefficient. In this article, we propose a high-performance and scalable traffic optimization strategy (HPSTOS) based on a hybrid approach that leverages the advantages of both centralized and distributed mechanisms. HPSTOS improves the efficiency of elephant flow detection through sampling and flow-table identification. HPSTOS guarantees preferential transmission of mice flows using priority scheduling and adjusts their transmission rate by coding-based congestion control on the end-host, reducing their latency. Additionally, HPSTOS schedules elephant flows by cost-aware dynamic flow scheduling on a centralized controller to improve their throughput. The controller handles only elephant flows, which constitutes the minority of the flows, allowing effective scalability. Evaluations show that HPSTOS outperforms existing schemes by realizing efficient elephant flow detection and improving network performance and scalability.

Recent advances in smart connected vehicles and Intelligent Transportation Systems (ITS) are based upon the capture and processing of large amounts of sensor data. Modern vehicles contain many internal sensors to monitor a wide range of mechanical and electrical systems and the move to semi-autonomous vehicles adds outward looking sensors such as cameras, lidar, and radar. ITS is starting to connect existing sensors such as road cameras, traffic density sensors, traffic speed sensors, emergency vehicle, and public transport transponders. This disparate range of data is then processed to produce a fused situation awareness of the road network and used to provide real-time management, with much of the decision making automated. Road networks have quiet periods followed by peak traffic periods and cloud computing can provide a good solution for dealing with peaks by providing offloading of processing and scaling-up as required, but in some situations latency to traditional cloud data centres is too high or bandwidth is too constrained. Cloud computing at the edge of the network, close to the vehicle and ITS sensor, can provide a solution for latency and bandwidth constraints but the high mobility of vehicles and heterogeneity of infrastructure still needs to be addressed. This paper surveys the literature for cloud computing use with ITS and connected vehicles and provides taxonomies for that plus their use cases. We finish by identifying where further research is needed in order to enable vehicles and ITS to use edge cloud computing in a fully managed and automated way. We surveyed 496 papers covering a seven-year timespan with the first paper appearing in 2013 and ending at the conclusion of 2019.

Data center networks demand high-performance, robust, and scalable load balancing protocols. Despite progress, existing work still cannot meet these requirements well. Software defined networking (SDN) can bring considerable flexibility to the management of data center networks. In the software-defined data center network, we design, analyze, and evaluate RSLB, a robust and scalable load balancing protocol that overcomes these challenges. RSLB uses fine-grained flowcell as the transmission unit, and uses link delay as the congestion metric. It uses a three-step routing strategy to route flowcells to the path with the least congestion. Through global congestion awareness, RSLB reduces flow completion time (FCT), and is more robust to topological asymmetries compared to existing congestion-agnostic schemes. To collect and store congestion information, RSLB adopts a distributed control structure that monitors the congestion of the entire network through multiple controllers, which makes it much more scalable for implementation in large-scale networks compare to existing congestion-aware schemes. The simulation results show that RSLB can achieve lower FCT for mice flows and higher throughput for elephant flows than existing schemes, no matter in failure-free topology or asymmetric topology.

Integrating Low Earth Orbit (LEO) satellites with terrestrial network infrastructures to support ubiquitous Internet service coverage has recently received increasing research momentum. One distinct challenge is the frequent topology change caused by the constellation behaviour of LEO satellites. In the context of software defined networking (SDN), the controller function that is originally required to control the conventional data plane fulfilled by terrestrial SDN switches will need to expand its responsibility to cover their counterparts in the space, namely LEO satellites that are used for data forwarding. As such, seamless integration of the fixed control plane on the ground and the mobile data plane fulfilled by constellation LEO satellites will become a distinct challenge. In this paper, we propose the Virtual Data-Plane Addressing (VDPA) Scheme by leveraging IP addresses to represent virtual switches at the fixed space locations which are periodically instantiated by the nested LEO satellites traversing them in a predictable manner. With such a scheme the changing data-plane network topology incurred by LEO satellite constellation can be made completely agnostic to the control plane on the ground, thus enabling a native approach to supporting seamless communication between the two planes. Our testbed-based experiment results prove the technical feasibility of the proposed VDPA-based flow rule manipulation mechanism in terms of data plane performance.

Software-defined networking (SDN) enables centralized control of a network of programmable switches by dynamically updating flow rules. This paves the way for dynamic and autonomous control of the network. In order to be able to apply a suitable set of policies to the correct set of traffic flows, SDN needs input from traffic classification mechanisms. Today, there is a variety of classification algorithms in machine learning. However, recent studies have found that using an arbitrary algorithm does not necessarily provide the best classification outcome on a dataset, and therefore a framework called ensemble which combines individual algorithms to improve classification results has gained attraction. In this paper, we propose the application of the ensemble algorithm as a machine learning pre-processing tool, which classifies ingress network traffic for SDN to pick the right set of traffic policies. Performance evaluation results show that this ensemble classifier can achieve robust performance in all tested traffic types.

Distributed Mobility Management (DMM) protocol is proposed to address the shortcomings of centralized mobility management protocols. In DMM, unlike centralized protocols, flows can be routed optimally in the network by dynamic IP address allocation, which can yield lower signaling and packet delivery costs. In this paper, we introduce an SDN based mobility management service with multiple controllers called Hierarchical Software Defined Distributed Mobility Management (HSD-DMM). In contrast to standard DMM which is proposed for flat architectures, HSD-DMM uses a dynamic anchor point selection method for each flow in a hierarchical mobile network architecture. The main condition in selecting an anchor point is packet delivery cost reduction based on collaboration of multiple controllers responsible for different tiers of the hierarchy. Numerical analysis results reveal that HSD-DMM can decrease signaling and packet delivery cost compared to an SDN based standard DMM solution.

Live holographic teleportation is an emerging media application that allows Internet users to communicate with each other in a fully immersive manner. One distinct feature of such an application is the capability of simultaneously teleporting multiple objects from different network locations to the receiver’s field of view, mimicking the effect of group-based communications in a common physical space. In this case, teleportation frames from individual sources need to be stringently synchronized in order to assure user Quality of Experiences (QoE) in terms of avoiding the perception of motion misalignment at the receiver side. In this paper, we carry out systematic performance evaluations on how different Internet path conditions may affect the teleportation frame synchronisation performances. Based on this, we present a lightweight, edge-computing based scheme that is able to achieve controllable frame synchronisation operations for multi-source based teleportation applications at the Internet scale.

Software-defined networking (SDN) is the key technology to enable network softwarization by offering a programable and flexible network control capabilities. In order to dynamically manage and reconfigure the underlying network through SDN, network-based monitoring functionality needs to be in place. However, existing network monitoring schemes are normally heavyweight which can cause substantial monitoring overhead when dealing with entire network infrastructure and complex policies. Such a limitation can be critical in a software-based network system that enables the construction of multiple networks with various network policies designed by a network operator. In this article, we propose a new lightweight monitoring mechanism referred to as Active-port Aware Monitoring (APAM) in order to support the monitoring of complex networks with substantially reduced overhead. APAM typically monitors active ports which are the switch ports utilized by current flow rules. These active ports are dynamically monitored with reconfigurable monitoring intervals according to their port utilization. The measurement results show that APAM adapts varying traffic route due to a change of flow rules and also adjusts its monitoring performance according to network traffic dynamicity, which reduces the monitoring overhead and also improves monitoring accuracy.

With the advent of Network Function Virtualization (NFV) techniques, a subset of the Internet traffic will be treated by a chain of virtual network functions (VNFs) during their journeys while the rest of the background traffic will still be carried based on traditional routing protocols. Under such a multi-service network environment, we consider the co-existence of heterogeneous traffic control mechanisms, including flexible, dynamic service function chaining (SFC) traffic control and static, dummy IP routing for the aforementioned two types of traffic that share common network resources. Depending on the traffic patterns of the background traffic which is statically routed through the traditional IP routing platform, we aim to perform dynamic service function chaining for the foreground traffic requiring VNF treatments, so that both the end-to-end SFC performance and the overall network resource utilization can be optimized. Towards this end, we propose a deep reinforcement learning based scheme to enable intelligent SFC routing decision-making in dynamic network conditions. The proposed scheme is ready to be deployed on both hybrid SDN/IP platforms and future advanced IP environments. Based on the real GEANT network topology and its one-week traffic traces, our experiments show that the proposed scheme is able to significantly improve from the traditional routing paradigm and achieve close-to-optimal performances very fast while satisfying the end-to-end SFC requirements.

The launch of the StarLink Project has recently stimulated a new wave of research on integrating Low Earth Orbit (LEO) satellite networks with the terrestrial Internet infrastructure. In this context, one distinct technical challenge to be tackled is the frequent topology change caused by the constellation behaviour of LEO satellites. Frequent change of the peering IP connection between the space and terrestrial Autonomous Systems (ASes) inevitably disrupts the Border Gateway Protocol (BGP) routing stability at the network boundaries which can be further propagated into the internal routing infrastructures within ASes. To tackle this problem, we introduce the Geosynchronous Network Grid Addressing (GNGA) scheme by decoupling IP addresses from physical network elements such as a LEO satellite. Specifically, according to the density of LEO satellites on the orbits, the IP addresses are allocated to a number of stationary "grids" in the sky and dynamically bound to the interfaces of the specific satellites moving into the grids along time. Such a scheme allows static peering connection between a terrestrial BGP speaker and a fixed external BGP (e-BGP) peer in the space, and hence is able to circumvent the exposure of routing disruptions to the legacy terrestrial ASes. This work-in-progress specifically addresses a number of fundamental technical issues pertaining to the design of the GNGA scheme.

With the high penetration of large-scale distributed renewable energy generation, the power system is facing enormous challenges in terms of the inherent uncertainty of power generation of renewable energy resources. In this regard, virtual power plants (VPPs) can play a crucial role in integrating a large number of distributed generation units (DGs) more effectively to improve the stability of the power systems. Due to the uncertainty and nonlinear characteristics of DGs, reliable economic dispatch in VPPs requires timely and reliable communication between DGs, and between the generation side and the load side. The online economic dispatch optimizes the cost of VPPs. In this article, we propose a deep reinforcement learning (DRL) algorithm for the optimal online economic dispatch strategy in VPPs. By utilizing DRL, our proposed algorithm reduced the computational complexity while also incorporating large and continuous state space due to the stochastic characteristics of distributed power generation. We further design an edge computing framework to handle the stochastic and large-state space characteristics of VPPs. The DRL-based real-time economic dispatch algorithm is executed online. We utilize real meteorological and load data to analyze and validate the performance of our proposed algorithm. The experimental results show that our proposed DRL-based algorithm can successfully learn the characteristics of DGs and industrial user demands. It can learn to choose actions to minimize the cost of VPPs. Compared with the deterministic policy gradient algorithm and DDPG, our proposed method has lower time complexity.

Network function virtualization (NFV) is one of key features envisioned for the upcoming 5G core networks in order to support high flexibility in network deployment and management. However, potential performance degradations that could be caused by virtualization of network functions is still a controversial issue, especially in regards to virtualization of core network components. In this paper, we evaluate the effect of NFV on an end-to-end mobile network testbed, which is deployed in the 5G Innovation Centre (5GIC) in University of Surrey. The testbed consists of indoor and outdoor LTE Radio Access Networks (RAN) equipment, as well as an enhanced Evolved Packet Core (EPC) following LTE Release 14 specifications, such as control and user plane separation (CUPS). The paper compares the performance of the softwarised core network and that of its virtualized counterpart. Measurement results show that the virtualized core network has adequately similar network performance in terms of throughput and latency, compared with the non-virtual core.

A techno‐economic analysis of the 5G enhanced mobile broadband scenario in dense urban areas has been accomplished by radio capacity modelling of probable 5G technologies within a 1 km 2 grid representing central London. Different density networks were modelled at 700 MHz (macro network), 3.5 GHz (micro network) and 24–27.5 GHz (hot spots) – together with 802.11ac access points. Using published data on network costs various deployment options have been evaluated for capacity, headline rate and capital expenditure/operating expense. It has been shown that reaching headline rates of 64–100 Mbps everywhere is possible with a number of different technology options. Massive increases in capacity (in excess of 100 Gbps/km 2 ), however, can only be realistically achieved with a millimetre wave (outdoor) and 802.11ac (internally). The cost of deploying such capacity, however, will be several times that of LTE – the authors estimate a 4–5 times increase in costs for a 100 Mbps everywhere network that has ×100 capacity increase over existing LTE networks. The cost of rolling out 5G is becoming an important issue and this work provides one of the few published estimates of the economics of ultra‐high capacity networks.

HTTP-based live streaming has become increasingly popular in recent years, and more users have started generating 4K live streams from their devices (e.g., mobile phones) through social-media service providers like Facebook or YouTube. If the audience is located far from a live stream source across the global Internet, TCP throughput becomes substantially suboptimal due to slow start and congestion control mechanisms. This is especially the case when the end-to-end content delivery path involves radio access network at the last mile. As a result, the data rate perceived by a mobile receiver may not meet the high requirement of 4K video streams, which causes deteriorated quality-of-experience (QoE). In this paper, we propose a scheme named edge-based transient holding of live segment (ETHLE), which addresses the above-mentioned issue by performing context-aware transient holding of video segments at the mobile edge with virtualized content caching capability. Through holding the minimum number of live video segments at the mobile edge cache in a context-aware manner, the ETHLE scheme is able to achieve seamless 4K live streaming experiences across the global Internet by eliminating buffering and substantially reducing initial startup delay and live stream latency. It has been deployed as a virtual network function at an LTE-A network, and its performance has been evaluated using real live stream sources that are distributed around the world. The significance of this paper is that by leveraging virtualized caching resources at the mobile edge, we address the conventional transport-layer bottleneck and enable QoE-assured Internet-wide live streaming services with high data rate requirements.

Software-defined networking (SDN) is a promising paradigm of computer networks, offering a programmable and centralized network architecture. However, although such a technology supports the ability to dynamically handle network traffic based on real-time and flexible traffic control, SDN-based networks can be vulnerable to dynamic change of flow control rules, which causes transmission disruption and packet loss in SDN hardware switches. This problem can be critical because the interruption and packet loss in SDN switches can bring additional performance degradation for SDN-controlled traffic flows in the data plane. In this paper, we propose a novel robust flow control mechanism referred to as priority-based flow control (PFC) for dynamic but disruption-free flow management when it is necessary to change flow control rules on the fly. PFC minimizes the complexity of flow modification process in SDN switches by temporarily adapting the priority of flow rules in order to substantially reduce the time spent on control-plane processing during run-time. Measurement results show that PFC is able to successfully prevent transmission disruption and packet loss events caused by traffic path changes, thus offering dynamic and lossless traffic control for SDN switches.

Due to dynamic wireless network conditions and heterogeneous mobile web content complexities, web-based content services in mobile network environments always suffer from long loading time. The new HTTP/2.0 protocol only adopts one single TCP connection, but recent research reveals that in real mobile environments, web downloading using single connection will experience long idle time and low bandwidth utilization, in particular with dynamic network conditions and web page characteristics. In this paper, by leveraging the Mobile Edge Computing (MEC) technique, we present the framework of Mobile Edge Hint (MEH), in order to enhance mobile web downloading performances. Specifically, the mobile edge collects and caches the meta-data of frequently visited web pages and also keeps monitoring the network conditions. Upon receiving requests on these popular webpages, the MEC server is able to hint back to the HTTP/2.0 clients on the optimized number of TCP connections that should be established for downloading the content. From the test results on real LTE testbed equipped with MEH, we observed up to 34.5% time reduction and in the median case the improvement is 20.5% compared to the plain over-the-top (OTT) HTTP/2.0 protocol.

User data allowance trading emerges as a promising practice in mobile data networks since it can help mobile networks to attract more users. However, to date, there is no study on user data allowance trading in mobile networks. In this paper, we develop a truthful framework that allows users to bid for data allowance. We focus on preventing price cheating, guaranteeing fairness, and minimizing trading maintenance cost in trading. We formulate the data trading process as a double auction problem and develop algorithms to solve the problem. In particular, we use a uniform price auction based on a competitive equilibrium to defend against price cheating and provide fair-ness. Meanwhile, we leverage linear programming to minimize trading maintenance cost. We conduct extensive simulations to demonstrate the performance of the proposed mechanism. The simulation results show that our trading mechanism is truthful and fair, while incurring a minimized maintenance cost.

In this paper, the design, construction, and testing of a fully-functional virtualised mobile core network is outlined. Lessons learned and recommendations for future improvements are provided. The presented work uses open-source software for virtual network function infrastructure control (OpenStack), network flow programming (OpenDaylight), and network orchestration (OpenBaton) to virtualise a commercial software evolved packet core solution deployed on common off-the-shelf hardware. The findings presented in this paper prove the concept of function virtualisation for mobile networks, and paves the way towards future mobile core network function flexibility as required for 5G networks. The paper provides researchers and network operators with first-hand experience to help build similar virtual mobile network infrastructures, and highlights the challenges to tackle and the issues to address to harness the power of virtualisation in 5G networks.

Internet video streaming applications have been demanding more bandwidth and higher video quality, especially with the advent of virtual reality and augmented reality appli-cations. While adaptive strea ming protocols like MPEG-DASH (dynamic adaptive streaming over HTTP) allows video quality to be flexibly adapted, e.g., degraded when mobile network condition deteriorates, this is not an option if the application itself requires guaranteed 4K quality at all time. On the other hand, conventional end-to-end transmission control protocol (TCP) has been struggling in supporting 4K video delivery across long-distance Internet paths containing both fixed and mobile network segments with heterogeneous characteristics. In this paper, we present a novel and practically feasible system architecture named MVP (mobile edge virtualization with adaptive prefetching), which enables content providers to embed their content intelligence as a virtual network function into the mobile network operator's infrastructure edge. Based on this architecture, we present a context-aware adaptive video prefetching scheme in order to achieve quality of experience (QoE)-assured 4K video on demand (VoD) delivery across the global Internet. Through experiments based on a real LTE-A network infrastructure, we demonstrate that our proposed scheme is able to achieve QoE-assured 4K VoD streaming, especially when the video source is located remotely in the public Internet, in which case none of the state-of-the-art solutions is able to support such an objective at global Internet scale.

The Internet-of-Things (IoT) paradigm envisions billions of devices all connected to the Internet, generating low-rate monitoring and measurement data to be delivered to application servers or end-users. Recently, the possibility of applying in-network data caching techniques to IoT traffic flows has been discussed in research forums. The main challenge as opposed to the typically cached content at routers, e.g., multimedia files, is that IoT data are transient and therefore require different caching policies. In fact, the emerging location-based services can also benefit from new caching techniques that are specifically designed for small transient data. This paper studies in-network caching of transient data at content routers, considering a key temporal data property: data item lifetime. An analytical model that captures the trade-off between multihop communication costs and data item freshness is proposed. Simulation results demonstrate that caching transient data are a promising information-centric networking technique that can reduce the distance between content requesters and the location in the network where the content is fetched from. To the best of our knowledge, this is a pioneering research work aiming to systematically analyze the feasibility and benefit of using Internet routers to cache transient data generated by IoT applications.

IP lookup plays a significant role in networking. The rapid development of the Internet brings new challenges to IP lookup in recent years. To deal with these challenges, we propose the first algorithm that we are aware of to use Minimal Perfect Hash (MPH) filters in IP lookup. It achieves the information theoretic optimum on-chip memory storage and O(1) worst case on-chip lookup speed. To overcome the shortcoming of MPH filter's no support for insertions, we propose an incremental update algorithm which achieves average update speed of O(1) memory access per update.

With the recent development of Device-to-Device (D2D) communication technologies, mobile devices will no longer be treated as pure “terminals”, but they could become an integral part of the network in specific application scenarios. In this paper, we introduce a novel scheme of using D2D communications for enabling data relay services in partial Not-Spots, where a client without local network access may require data relay by other devices. Depending on specific social application scenarios that can leverage on the D2D technology, we consider tailored algorithms in order to achieve optimised data relay service performance on top of our proposed network-coordinated communication framework. The approach is to exploit the network's knowledge on its local user mobility patterns in order to identify best helper devices participating in data relay operations. This framework also comes with our proposed helper selection optimization algorithm based on reactive predictability of individual user. According to our simulation analysis based on both theoretical mobility models and real human mobility data traces, the proposed scheme is able to flexibly support different service requirements in specific social application scenarios.

User data allowance trading is emerging as a promising field in mobile data networks. Mobile operators are establishing data trading platforms to attract more users. To date, there has been no coherent study on user data allowance trading. In this paper, we develop a truthful framework that allows users to bid for data allowance. We focus on preventing price cheating, guaranteeing fairness and minimizing trading maintenance cost. We model the data trading process as a double auction problem. We develop algorithms to solve the problem. The algorithms use a uniform price based on a competitive equilibrium to defend against price cheating and provide fairness, and use linear programming to minimize trading maintenance cost. We conduct extensive simulations to testify the proposed mechanism. Results show that our mechanism is truthful, fair and can minimize the cost of trading.

The size of Forwarding Information Base (FIB) maintained at backbone routers is experiencing an exponential growth, and various solutions have been proposed in the literature. The main shortcoming of FIB compression is the update overhead. Only when the update speed of FIB compression algorithms is sufficiently fast and bounded, the probability of packet loss incurred by FIB compression operations during update can be completely avoided. However, no prior FIB compression algorithm can bound the worst case of update, and hence a mature solution with complete avoidance of packet loss is still yet to be identified. To address this issue, we propose the Unite and Split (US) compression algorithm to enable fast update with bounded worst case performance. Experimental results show that the average update speed of the US algorithm is almost the same as that of the binary trie without any compression.

With the advent of device-to-device (D2D) communications, user equipment (UE) such as smart phones will become an integral part of the (mobile) network for content distribution operations. In this context, we introduce a novel Information centric networking (ICN) framework for mobile content caching and distribution based on direct D2D communications in cellular network environments. Specifically, small pieces of mobile content can be cached at incentivised mobile UEs known as helpers, and the ICN-aware cellular network edge (e.g. base stations) are able to resolve content requests from local mobile clients to the helpers in their D2D proximity. With simple POI (point of interests) based selection of helpers as well as content caching/eviction control at the base station side, a significant proportion of mobile content requests can be locally resolved to helpers in proximity of clients, thus achieving very effective content traffic offloading away from the cellular network infrastructure.

Node clustering has been widely studied in recent years for Wireless Sensor Networks (WSN) as a technique to form a hierarchical structure and prolong network lifetime by reducing the number of packet transmissions. Cluster Heads (CH) are elected in a distributed way among sensors, but are often highly overloaded, and therefore re-clustering operations should be performed to share the resource intensive CH-role. Existing protocols involve periodic network-wide re-clustering operations that are simultaneously performed, which requires global time synchronisation. To address this issue, some recent studies have proposed asynchronous node clustering for networks with direct links from CHs to the data sink. However, for large-scale WSNs, multihop packet delivery to the sink is required since longrange transmissions are costly for sensor nodes. In this paper, we present an asynchronous node clustering protocol designed for multihop WSNs, considering dynamic conditions such as residual node energy levels and unbalanced data traffic loads caused by packet forwarding. Simulation results demonstrate that it is possible to achieve similar levels of lifetime extension by re-clustering a multihop WSN via independently made decisions at CHs, without a need for time synchronisation required by existing synchronous protocols.

The recent forecast of billions of devices, all connected to the Internet and generating low-rate monitoring, measurement, or automation data that many end-users/applications frequently request, signifies the need for applying in-network caching techniques to Internet-of-Things (IoT) traffic. Although time delay is not critically important for small-sized IoT content, the expected total traffic load on the Internet from a large number of devices is significant. However, the main challenge as opposed to the typically cached content at content routers, e.g. multimedia files, is that IoT data are transient and therefore require different caching policies. This paper studies in-network caching of IoT data at content routers in the Internet. An IoT data item is uniquely defined not only by its time and location tags, but also a time-range value set by end-users/applications. We provide a model for the trade-off between multihop communication costs and the freshness of a transient data item. Results show that the model can successfully capture the effect of data transiency and can accurately represent the expected gains of a caching system: considerable savings in terms of reduction of network load, especially for highly requested data items.

Reducing energy consumption in the Telecom industry has become a major research challenge to the Internet community. Towards this end, numerous research works have been carried out to mitigate the growth of energy consumption through intelligent network control mechanisms. This paper proposes a novel approach to achieving energy efficiency in ISP backbone networks according to dynamic traffic conditions. The main objective is to enforce as many links as possible to go to sleep during the off-peak time, while in event of traffic volume increase, the minimum number of sleeping links should be required to wake up to handle this dynamicity and in a way that this creates minimal or no traffic disruption. Based on our simulations with the GEANT and Abilene network topologies and their traffic traces respectively, up to 47% and 44% energy gains can be achieved without any obstruction to the network performance. Secondly, we show that the activation of a small number of sleeping links is still sufficient to cope with any traffic surge instead of reverting to the full topology or sacrificing energy savings as seen in some research proposals.

Backup paths are usually pre-installed by network operators to protect against single link failures in backbone networks that use multi-protocol label switching. This paper introduces a new scheme called Green Backup Paths (GBP) that intelligently exploits these existing backup paths to perform energy-aware traffic engineering without adversely impacting the primary role of these backup paths of preventing traffic loss upon single link failures. This is in sharp contrast to most existing schemes that tackle energy efficiency and link failure protection separately, resulting in substantially high operational costs. GBP works in an online and distributed fashion, where each router periodically monitors its local traffic conditions and cooperatively determines how to reroute traffic so that the highest number of physical links can go to sleep for energy saving. Furthermore, our approach maintains quality-of-service by restricting the use of long backup paths for failure protection only, and therefore, GBP avoids substantially increased packet delays. GBP was evaluated on the point-of-presence representation of two publicly available network topologies, namely, GÉANT and Abilene, and their real traffic matrices. GBP was able to achieve significant energy saving gains, which are always within 15% of the theoretical upper bound.

In a content delivery network (CDN), the energy cost is dominated by its geographically distributed data centers (DCs). Generally within a DC, the energy consumption is dominated by its server infrastructure and cooling system, with each contributing approximately half. However, existing research work has been addressing energy efficiency on these two sides separately. In this paper, we jointly optimize the energy consumption of both server infrastructures and cooling systems in a holistic manner. Such an objective is achieved through both strategies of: 1) putting idle servers to sleep within individual DCs; and 2) shutting down idle DCs entirely during off-peak hours. Based on these strategies, we develop a heuristic algorithm, which concentrates user request resolution to fewer DCs, so that some DCs may become completely idle and hence have the opportunity to be shut down to reduce their cooling energy consumption. Meanwhile, QoS constraints are respected in the algorithm to assure service availability and end-to-end delay. Through simulations under realistic scenarios, our algorithm is able to achieve an energy-saving gain of up to 62.1% over an existing CDN energy-saving scheme. This result is bound to be near-optimal by our theoretically-derived lower bound on energy-saving performance.

Energy constraint is a crucial factor in wireless sensor networks (WSNs). One of the best solutions for networking in the WSNs is a clustering-based technique to deal with the energy constraint. However, cluster heads (CHs) in such technique tend to consume much more energy than ordinary nodes and, eventually, deplete quickly. In this paper, we propose a clustering technique, included CH selection and rotation, using an event-driven data reporting during continuous data monitoring of ambient. SNs in this technique report only necessary data when data changes exceeding a given threshold. Therefore, clusters are created only upon specific places where such necessary data changes are happening. Furthermore, the clusters are operated as long as the ambient situation is changing. Once the situation becomes stable, the clusters will be reset and every sensor node in these clusters switch to sleep mode in order to conserve energy consumed by CHs and members. Results show that the network lifetime and stability is better than some existing protocols.

This paper presents a holistic peer selection scheme in multi-domain environments, aiming to mitigate Peer-to-Peer (P2P) traffic volumes over expensive inter-domain links as well as the maintenance of desirable P2P users' perceived service quality. The mechanism combines the traditional locality-aware peer selection with the consideration of ISP business relationship. By leveraging between the two peering strategies, the risk of possible congestion on critical inter-connected links can be effectively alleviated due to more concentrated P2P traffic over fewer inter-ISP links under pure cooperative peering schemes. According to our analytical modelling, the proposed hybrid approach is able to achieve better performance for P2P users, and can retain desirable network efficiency as of the cooperative peer selection strategy. Our modelling based analysis offers the incentives to perform peer selections in multi-domain environments wherein non-cooperative networks and cooperative networks coexist.

With the popularity of information and content items that can be cached within ISP networks, developing high-quality and efficient content distribution approaches has become an important task in future internet architecture design. As one of the main techniques of content distribution, in-network caching mechanism has attracted attention from both academia and industry. However, the general evaluation model of in-network caching is seldom discussed. The trade-off between economic cost and the deployment of in-network caching still remains largely unclear, especially for heterogeneous applications. We take a first yet important step towards the design of a better evaluation model based on the Application Adaptation CapaciTy (2ACT) of the architecture to quantify the trade-off in this paper. Based on our evaluation model, we further clarify the deployment requirements for the in-network caching mechanism. Based on our findings, ISPs and users can make their own choice according to their application scenarios. © 2013 IEEE.

Energy consumption in ISP backbone networks has been rapidly increasing with the advent of increasingly bandwidth-hungry applications. Network resource optimization through sleeping reconfiguration and rate adaptation has been proposed for reducing energy consumption when the traffic demands are at their low levels. It has been observed that many operational backbone networks exhibit regular diurnal traffic patterns, which offers the opportunity to apply simple time-driven link sleeping reconfigurations for energy-saving purposes. In this work, an efficient optimization scheme called Time-driven Link Sleeping (TLS) is proposed for practical energy management which produces an optimized combination of the reduced network topology and its unified off-peak configuration duration in daily operations. Such a scheme significantly eases the operational complexity at the ISP side for energy saving, but without resorting to complicated online network adaptations. The GÉANT network and its real traffic matrices were used to evaluate the proposed TLS scheme. Simulation results show that up to 28.3% energy savings can be achieved during off-peak operation without network performance deterioration. In addition, considering the potential risk of traffic congestion caused by unexpected network failures based on the reduced topology during off-peak time, we further propose a robust TLS scheme with Single Link Failure Protection (TLS-SLFP) which aims to achieve an optimized trade-off between network robustness and energy efficiency performance.

Receiving great interest from the research community, Delay Tolerant Networks (DTNs) are a type of Next Generation Networks (NGNs) proposed to bridge communication in challenged environments. In this paper, the message replication probability is proportionally sprayed for efficient routing mainly under sparse scenario. This methodology is different from the spray based algorithms using message copy tickets to control replication. Our heuristic algorithm aims to overcome the scalability of the spray based algorithms, since to determine the initial value of the copy tickets requires the assumption that either the number of nodes is known in advance, or the underlying mobility model follows the Random WayPoint (RWP) characteristic. Specifically, in combining with the assistance of geographic information to estimate the movement range of destination, the routing decision is based on the encounter angle between pairwise nodes, and is dynamically switched between the designed two routing phases, named as geographic replication and replication probability spray. Furthermore, messages are under prioritized transmission with the consideration of redundancy pruning. Simulation results show our heuristic algorithm outperforms other well known algorithms in terms of delivery ratio, transmission overhead, average latency as well as buffer occupancy time.

Optimizing server's power consumption in content distribution infrastructure has attracted increasing research efforts. The technical challenge is the tradeoff between server power consumption and the content service capability on both the server and the network side. This paper proposes and evaluates a novel approach that optimizes content servers' power consumptions in large-scale content distribution platforms across multiple ISP domains. Specifically, our approach strategically puts servers to sleep mode without violating load capacities of virtual content delivery links and active servers in the infrastructure. Such a problem can be formulated into a nonlinear programming model. The efficiency of our approach is evaluated in a content distribution topology covering two real interconnected domains. The simulation has shown that our approach is capable of reducing servers' power consumptions by up to 62.2%, while maintaining the actual service performance in an acceptable scope.

The continuous growth in volume of Internet traffic, including VoIP, IPTV and user-generated content, requires improved routing mechanisms that satisfy the requirements of both the Internet Service Providers (ISPs) that manage the network and the end-users that are the sources and sinks of data. The objectives of these two players are different, since ISPs are typically interested in ensuring optimised network utilisation and high throughput whereas end-users might require a low-delay or a high-bandwidth path. In this paper, we present our UAESR (Utilisation-Aware Edge Selected Routing) algorithm, which aims to satisfy both players' demands concurrently by selecting paths that are a good compromise between the two players' objectives. We demonstrate by simulation that this algorithm allows both actors achieve their goals. The results support our argument that our cooperative approach achieves effective network resource engineering at the same time as offering routing flexibility and good quality of service to end-users.

Backbone network energy efficiency has recently become a primary concern for Internet Service Providers and regulators. The common solutions for energy conservation in such an environment include sleep mode reconfigurations and rate adaptation at network devices when the traffic volume is low. It has been observed that many ISP networks exhibit regular traffic dynamicity patterns which can be exploited for practical time-driven link sleeping configurations. In this work, we propose a joint optimization algorithm to compute the reduced network topology and its actual configuration duration during daily operations. The main idea is first to intelligently remove network links using a greedy heuristic, without causing network congestion during off-peak time. Following that, a robust algorithm is applied to determine the window size of the configuration duration of the reduced topology, making sure that a unified configuration with optimized energy efficiency performance can be enforced exactly at the same time period on a daily basis. Our algorithm was evaluated using on a Point-of-Presence representation of the GÉANT network and its real traffic matrices. According to our simulation results, the reduced network topology obtained is able to achieve 18.6% energy reduction during that period without causing significant network performance deterioration. The contribution from this work is a practical but efficient approach for energy savings in ISP networks, which can be directly deployed on legacy routing platforms without requiring any protocol extension.

In order to meet the requirements of emerging demanding services, network resource management functionality that is decentralized, flexible and adaptive to traffic and network dynamics is of paramount importance. In this paper we describe the main mechanisms of DACoRM, a new intra-domain adaptive resource management approach for IP networks. Based on path diversity provided by multi-topology routing, our approach controls the distribution of traffic load in the network in an adaptive manner through periodical re-configurations that uses real-time monitoring information. The re-configuration actions performed are decided in a coordinated fashion between a set of source nodes that form an in-network overlay. We evaluate the overall performance of our approach using realistic network topologies. Results show that near-optimal network performance in terms of resource utilization can be achieved in scalable manner.

Without the assumption of contemporaneous end to end connectivity in challenged wireless networks, Delay Tolerant Networking (DTN) routing is an important research area. The contribution in this paper is to take advantage of the proposed DTN geographic replication to overcome the limitation of topology based utility replication, since message replication is prevented due to the local maximum problem that the utility metric of encountered node is worse than message carrier. In brief, the proposed DTN geographic replication is activated only if the utility replication is unable to route message, this hybrid approach promotes a seamless message replication given limited message lifetime. Borrowing from the concept of gravity, messages are under prioritized transmission for load balancing and achieving less delivery latency. Extensive simulation results show promising improvement of the proposed algorithm in terms of delivery ratio, transmission cost, average latency as well as number of aborted messages.

Network virtualization has been recognized as a promising solution to enable the rapid deployment of customized services by building multiple Virtual Networks (VNs) on a shared substrate network. Whereas various VN embedding schemes have been proposed to allocate the substrate resources to each VN requests, little work has been done to provide backup mechanisms in case of substrate network failures. In a virtualized infrastructure, a single substrate failure will affect all the VNs sharing that resource. Provisioning a dedicated backup network for each VN is not efficient in terms of substrate resource utilization. In this paper, we investigate the problem of shared backup network provision for VN embedding and propose two schemes: shared on-demand and shared pre-allocation backup schemes. Simulation experiments show that both proposed schemes make better utilization of substrate resources than the dedicated backup scheme without sharing, while each of them has its own advantages.

Fast reroute (FRR) techniques have been designed and standardised in recent years for supporting sub-50-millisecond failure recovery in operational ISP networks. On the other hand, if the provisioning of FRR protection paths does not take into account traffic engineering (TE) requirements, customer traffic may still get disrupted due to post-failure traffic congestion. Such a situation could be more severe in operational networks with highly dynamic traffic patterns. In this paper we propose a distributed technique that enables adaptive control of FRR protection paths against dynamic traffic conditions, resulting in self-optimisation in addition to the self-healing capability. Our approach is based on the Loop-free Alternates (LFA) mechanism that allows non-deterministic provisioning of protection paths. The idea is for repairing routers to periodically re-compute LFA alternative next-hops using a lightweight algorithm for achieving and maintaining optimised post-failure traffic distribution in dynamic network environments. Our experiments based on a real operational network topology and traffic traces across 24 hours have shown that such an approach is able to significantly enhance relevant network performance compared to both TE-agnostic and static TE-aware FRR solutions.

The recently proposed Application Layer Traffic Optimization (ALTO) framework has opened up a new dimension for Internet traffic management that is complementary to the traditional application-agnostic traffic engineering (AATE) solutions currently employed by ISPs. In this paper, we investigate how ALTO-assisted Peer-to-Peer (P2P) traffic management functions interact with the underlying AATE operations, given that there may exist different application-layer policies in the P2P overlay. By considering specific P2P peer selection behaviors on top of a traffic-engineered ISP network, we conduct a performance analysis on how the application and network-layer respective performance is influenced by different policies at the P2P side. Our empirical study offers significant insight for the future design and analysis of cross-layer network engineering approaches that involve multiple autonomous optimization entities with both consistent and non-consistent policies.

In this paper we introduce a new scheme to achieve fast failure recovery in IP multicast based content delivery, which is based on efficient extensions to the Not-via fast reroute (FRR) technique. The design of such an approach takes into account distinct characteristics of IP multicast routing, namely receiver-initiated and state-based, and it offers comprehensive protections against both simple and complex network failures. We also specify in the paper moderate extensions to the standard PIM-SM routing protocol in order to equip individual repairing routers with necessary knowledge for dynamically binding protected multicast trees with pre-established Not-via tunnels that are able to automatically bypass failed network components. Our simulation experiments based on both real and synthetically generated topologies indicate promising scalability performance in the proposed multicast FRR approach.

Emerging Peer-to-Peer (P2P) technologies have enabled various types of content to be efficiently distributed over the Internet. Most P2P systems adopt selfish peer selection schemes in the application layer that in some sense optimize the user quality of experience. On the network side, traffic engineering (TE) is deployed by ISPs in order to achieve overall efficient network resource utilization. These TE operations are typically performed without distinguishing between P2P flows and other types of traffic. Due to inconsistent or even conflicting objectives from the perspectives of P2P overlay and network-level TE, the interactions between the two and their impact on the performance for each is likely to be non-optimal, and also has not yet been investigated in detail. In this paper we study such non-cooperative interactions by modeling best-reply dynamics, in which the P2P overlay and network-level TE optimize their own strategies based on the decision of the other player in the previous round. According to our simulations results based on data from the ABILENE network, P2P overlays exhibit strong resilience to adverse TE operations in maintaining end-to-end performance at the application layer. In addition, we show that network-level TE may suffer from performance deterioration caused by greedy peer (re-)selection behavior in reacting to previous TE adjustments.

With the increasing importance of the Internet for delivering personal and business applications, the slow re-convergence after network failure of existing routing protocols becomes a significant problem. This is especially true for real time multimedia services where service disruption cannot be generally tolerated. In order to ensure fast network failure recovery, IP Fast Reroute (FRR) can be adopted to immediately reroute affected customer traffic from the default path onto a backup path when link failure occurs, thus avoiding slow Interior Gateway Protocol (IGP) re-convergence. We notice that IGP link weight setting plays an important role in influencing the protection coverage performance in intra-domain link failures. Therefore in this paper we present an IGP link weight optimization scheme for backup path provisioning, which works on top of a multi-plane enabled routing platform. The scheme aims to optimize the path diversity among multiple routing planes. Due to the large search space of possible intra-domain link weights, in this paper we adopted a global search method based on a Genetic Algorithm to optimize the IGP link weights. Evaluation results show that in most cases a set of optimal link weights can be found which ensures that there are no more critical shared links among all the diverse paths on each routing plane. As a result, backup paths can be always available in case of single link failures.

Exploiting path diversity to enhance communication reliability is a key desired property in Internet. While the existing routing architecture is reluctant to adopt changes, overlay routing has been proposed to circumvent the constraints of native routing by employing intermediary relays. However, the selfish inter-domain relay placement may violate local routing policies at intermediary relays and thus affect their economic costs and performances. With the recent advance of the concept of network virtualization, it is envisioned that virtual networks should be provisioned in cooperation with infrastructure providers in a holistic view without compromising their profits. In this paper, the problem of policy-aware virtual relay placement is first studied to investigate the feasibility of provisioning policy-compliant multipath routing via virtual relays for inter-domain communication reliability. By evaluation on a real domain-level Internet topology, it is demonstrated that policy-compliant virtual relaying can achieve a similar protection gain against single link failures compared to its selfish counterpart. It is also shown that the presented heuristic placement strategies perform well to approach the optimal solution.

As a scalable paradigm for content distribution at Internet-wide scale, Peer-to-Peer (P2P) technologies have enabled a variety of networked services, such as distributed file-sharing and live video streaming. Most existing P2P systems employ non-intelligent peer selection algorithms for content swarming which greedily consume Internet bandwidth resources. As a result, Internet service providers (ISPs) need some efficient solutions for managing P2P traffic within their own networks. A common practice today is to block or shape P2P traffic in order to conserve bandwidth resources for carrying standard traffic from which revenue can be generated. In this paper, instead of looking at simple time-driven blocking/limiting approaches, we investigate how such types of limiting behaviors can be more gracefully performed by the ISP by taking into account the dynamics of both P2P traffic and of standard Internet traffic. Specifically, our approach is to adaptively limit excessive P2P traffic on critical network links that are prone to congestion, based on periodical link load/utilization measurements by the ISP. The ultimate objective is to guarantee non-P2P service capability while trying to accommodate as much P2P traffic as possible based on the available bandwidth resources. This approach can be regarded as a complementary solution to the recently proposed collaboration-based P2P paradigms such as P4P. Simulation results show that our approach not only eliminates performance degradation of non-P2P services that are caused by overwhelming P2P traffic, but also accommodates P2P traffic efficiently in both existing and future collaboration-based P2P network scenarios.

We present an efficient multi-plane based fast network failure recovery scheme which can be realized using the recently proposed multi-path enabled BGP platforms. We mainly focus on the recovery scheme that takes into account BGP routing disruption avoidance at network boundaries, which can be caused by intra-AS failures due to the hot potato routing effect. On top of this scheme, an intelligent IP crank-back operation is also introduced for further enhancement of network protection capability against failures. Our simulations based on both real operational network topologies and synthetically generated ones suggest that, through our proposed optimized backup egress point selection algorithm, as few as two routing planes are able to achieve high degree of path diversity for fast recovery in any single link failure scenario.

Locality-based peer selection paradigms have been proposed recently based on cooperation between peer-to-peer (P2P) service providers, Internet Service Providers (ISPs) and end users in order to achieve efficient resource utilization by P2P traffic. Based on this cooperation between different stakeholders, we introduce a more advanced paradigm with adaptive peer selection that takes into account traffic dynamics in the operational network. Specifically, peers associated with low path utilization as measured by the ISP are selected in order to reduce the probability of network congestion. This approach not only improves real-time P2P service assurance but also optimizes the overall use of network resources. Our simulations based on the GEANT network topology and real traffic traces show that the proposed adaptive peer selection scheme achieves significant improvement in utilizing bandwidth resources as compared to static locality-based approaches.

Traffic Engineering (TE) involves network configuration in order to achieve optimal IP network performance. The existing literature considers intra- and inter-AS (Autonomous System) TE independently. However, if these two aspects are considered separately, the overall network performance may not be truly optimized. This is due to the interaction between intra and inter-AS TE, where a good solution of inter-AS TE may not be good for intra-AS TE. To remedy this situation, we propose a joint optimization of intra- and inter-AS TE in order to improve the overall network performance by simultaneously finding the best egress points for inter-AS traffic and the best routing scheme for intra-AS traffic. Three strategies are presented to attack the problem, sequential, nested and integrated optimization. Our evaluation shows that, in comparison to sequential and nested optimization, integrated optimization can significantly improve overall network performance by being able to accommodate approximately 30%-60% more traffic demand.

Intra- and inter-AS transient link failures are common in operational IP networks. Robust intra- and inter-AS traffic engineering (TE) schemes have been proposed to optimize network performance against transient link failures. The existing literature has focused solely on either intra- or inter-AS link failure. They have, however, neglected the interactions between robust intra- and inter-AS TE, specifically the impact of intra-AS link failure on inter-AS TE and vice versa. As a result, the overall network performance may not be truly robust to link failures if the interactions are neglected. This paper proposes a joint robust TE approach that takes the interactions into account for achieving good network performance under both normal state and any single intra- or inter-AS link failure. We propose a two-phase heuristic to solve the problem and compare its performance with four alternative approaches that do not consider the interactions. Evaluation results reveal that our joint robust TE approach achieves higher robustness against intra- and inter-AS link failures than all the alternatives.

IP Fast ReRoute (FRR) mechanisms have been proposed to achieve fast failover for supporting Quality of Services (QoS) assurance. However, these mechanisms do not consider network performance after affected traffic is rerouted onto repair paths. As a result, QoS deterioration may still happen due to

In this paper we introduce the concepts of Network Planes and Parallel Internets, with the objective of designing and implementing a lightweight solution for viable end-to-end QoS provisioning. The proposed solution can be deployed with very small incremental additions to the existing best-effort Internet. Through Network Plane engineering and interconnection, mainly by means of intra- and inter-domain routing differentiation, end-to-end service differentiation across the Internet can be achieved.

Multicast traffic engineering (TE) has recently attracted significant attention given the emergence of point-to-multipoint multimedia content delivery over the Internet. Existing multicast resource provisioning solutions tend to use explicit-routing based TE with multiprotocol label switching (MPLS) tunnels. In this paper, we shift away from this overlay approach and address native IP multicast traffic engineering based on link state routing protocols. The objective is that, through plain protocol independent multicast-sparse mode (PIM-SM) shortest path routing with optimized multitopology IGP (MT-IGP) link weights, the resulting multicast trees are geared towards minimal consumption of bandwidth resources. We apply genetic algorithms (GA) to the calculation of optimized MT-IGP link weights that specifically cater for engineered PIM-SM routing with statistical bandwidth guarantees in multimedia content delivery. Our evaluation results show that GA-based multicast traffic engineering consumes significantly less bandwidth in comparison to conventional IP approaches while also exhibiting higher service availability

Traffic Engineering (TE) is used to optimize IP operational network performance. The existing literature generally considers intra- and inter-AS (Autonomous System) TE independently. However, the overall network performance may not be truly optimized when these aspects are considered separately. This is due to the interaction between intra- and inter-AS TE, where a solution of intra-AS TE may not be a good input to inter-AS TE and vice versa. To remedy this situation, we propose considering intra-AS aspects during inter-AS TE and vice versa. We propose a joint optimization of intra- and inter-AS TE to further improve the overall network performance by simultaneously finding the best egress points for the inter-AS traffic and the best routing scheme for the intra-AS traffic. Three strategies are presented to attack the problem, namely sequential, nested and integrated optimization. Our simulation study shows that, compared to sequential and nested optimization, integrated optimization can significantly improve the overall network performance by accommodating 30%-60% more traffic demands.

Bi-directional shared tree is an efficient routing scheme for many-to-many multicast applications (e.g. multiparty videoconferencing, interactive distance lecturing and Internet games etc). Given the open-group IP multicast service model, it is important to perform sender access control so as to prevent group members from receiving irrelevant data, and also protect the multicast tree from various Denial-of-Service (DoS) attacks. In comparison to source based and unidirectional shared trees where the data source can be authorized or authenticated at the single root or rendezvous point, in bi-directional routing this is a much more difficult problem since hosts can send data to all group members directly from any point in the tree. In this paper we propose a dynamic sender access control mechanism for bi-directional multicast trees so that irrelevant data is policed and discarded as it reaches any on-tree router. We show through simulation that the overhead of our mechanism is relatively small in terms of required state information in routers so that the proposed approach scales well for large groups.