DiffServ Networks Research Articles

A new framework for MPEG video delivery over heterogeneous networks

The failure of legacy packet markers, such as the Single Rate Three Color Marker (SRTCM) and the Two Rate Three Color Marker (TRTCM), to protect important data from being dropped, renders a poor delivery of video quality over Differentiated Services (DiffServ) networks due to the hierarchical structure of MPEG video. Therefore, we propose a content-aware packet marking scheme, which we call Two Markers System (TMS), to improve the delivery quality of MPEG video streams. TMS places marker modules at both the video source and the edge of a DiffServ network. The source marking modules act at the application-level to set an importance indicator in the IP header of the MPEG frame packet by evaluating the contribution of the packet to the expected picture quality perceived by an end user. Meanwhile, at the edge of the DiffServ network, the Enhanced Token Bucket Three Color Marker (ETBTCM) is proposed to perform packet marking at the network-level in accordance with the pre-assigned importance of individual video frame packets. When the traffic flow is measured as out-of-profile, the less important packets will be marked with higher probability than the more important ones. Therefore, in the event of network congestion, less important frame packets can be dropped before more important ones. Consequently, a better delivered video quality can be provided. Further, due to the convenience offered by wireless networks, more users choose mobile devices to connect to the Internet. Therefore, a Video Differentiator is suggested, placed at the entry to the IEEE 802.11e wireless network, to provide traffic differentiation and better service for video traffic. We use authentic MPEG4 video traffic traces to evaluate this framework. Results show that better video quality can be delivered over our proposed new framework.

Adaptive QoS provisioning by pricing incentive QoS routing for next generation networks

In this paper, we present an integrated solution for enabling the next generation Internet to achieve the differentiated service and availability guarantee. In particular, we utilize a recent proposed service architecture, referred to as Service Vector (SV), to enhance the QoS granularity and flexibility offered in the Diffserv network model and improve both the network resource utilization and user benefits. In order to efficiently and effectively use network resources, we investigate the issue of integrating pricing into QoS routing and propose a PRicing InCEntive QoS Routing (PRICER) mechanism. PRICER consists of two components: a novel routing-oriented state update (ROSE) scheme and an efficient Pricing Incentive Routing Algorithm (PIRA). ROSE performs the task of exchanging link state information throughout the network, and PIRA is a routing algorithm used to find paths meeting the QoS requirements of applications. By theoretical analysis and extensive simulations, we show that ROSE greatly outperforms the state of the arts in terms of both protocol overhead and the accuracy of link state information, and PIRA can guarantee finding the QoS constrained path with fairly low average computational complexity. The most distinguished property of PIRA is its progressive property, which is very useful in practice: it can self-adaptively minimize its computational complexity without sacrificing its performance. Another contribution of this paper is the introduction of a method to numerically evaluate the staleness of link state information.

Statistical delay control scheme for DiffServ networks with self-similar traffic

A statistical traffic control scheme is proposed, which computes the expected queue-length boundary conditions for self-similar traffic priority classes and can adapt the traffic class service level to provide practical and effective guaranteed quality of service control support in DiffServ networks.

Influence of Sender Parameters and Network Architecture on Perceived Audio Quality

Network simulations and subjective testing have been used to evaluate the effects of frame lengths, packetization settings, and the advantages of using a Differentiated Services (DiffServ) network architecture for music streaming over Internet Protocol (IP) networks. The simulations show that using smaller packet sizes can result in lower packet loss ratios, even though the total bit rate is increased due to network overhead. The network architecture simulations show that DiffServ make the packet loss process less bursty when compared to the current Best Effort (BE) network architecture. These network simulations were followed up with a subjective test to investigate the effects more thoroughly. The subjective test combined three audio clips, three packet loss ratios (3%, 5%, 7%) and two Error Concealment (EC) algorithms with two packet loss processes. Results from the subjective test show statistically significant effects of audio clip, EC, and packet loss ratio on perceived audio quality as well as interaction effects between the network architectures and both audio clip and packet loss ratio. It was found that for a loss ratio of 3%, the loss pattern from a DiffServ network lead to a significantly higher perceived audio quality than the loss pattern from a BE network. However, for higher loss ratios no differences can be seen between the two network architectures.

Edge overlay multicast to support comparable multi-class services

Traditional IP multicast is likely to imply a huge burden of storage and forwarding for routers. Some issues, such as the scalability problem, will be aggravated when various classes of quality of service are fulfilled in multicast traffic. The recent proposed application layer multicast is more scalable but increases traffic load in network layers and end-to-end delivery delay. In this paper, we present a multicast protocol that supports comparable multi-class services over an overlay network where edge routers are multicast-aware and core routers are oblivious of multicast. The comparable multi-class services consist of different services required by multicast members which can be satisfied via the same delivery tree. Considering the multi-class services and resource limitation on routers, the problem of building minimum cost trees has been proven to be NP-hard. So, we design several feasible heuristic algorithms to solve it. Extensive simulations are conducted to evaluate the performance of the proposed heuristics, and validate the effectiveness of reducing the total tree cost and iteration times under considered constraints. The proposed method is expected to be combined with DiffServ or MPLS VPN networks to fulfill multi-class QoS overlay multicast.

Adaptive Sliding Mode Congestion Control for DiffServ Network

This paper is about the application of adaptive sliding mode control to solving congestion problem for DiffServ Network. A nonlinear fluid flow model which is transformed into a parametric strict feedback form is used to analyze and control. Backstepping design procedure is applied, which leads to a new adaptive sliding mode controller. And by using the adaptive sliding mode controller, we achieve buffer queue length regulation against model uncertainties and disturbances. The simulation results show that the proposed control system is robust to unknown nonlinearities and disturbances.

Robust AQM Controller Design for DiffServ Network Using Sliding Mode Control

For DiffServ network, an AQM controller is designed based on the nonlinear fluid flow model and the Integrated Dynamic Congestion Control (IDCC) scheme. Due to modeling uncertainties, time varying parameter fluctuations and external disturbances, we adopt a robust controller based on sliding mode control. The controller is designed for premium traffic service and ordinary traffic service respectively. Especially for ordinary traffic, a sliding mode control algorithm is designed to compensate the input delay. Simulation results demonstrate that this method enables the queue length trace the reference value quickly. Consequently the proposed scheme gets good control effect.

Enhancement of fairness in a DiffServ network using a novel queuing algorithm

Core-Stateless Fair Queuing (CSFQ) algorithm was proposed to allocate bandwidth fairly at routers in the Differentiated Services Model (DiffServ Model). However, CSFQ can not effectively provide the fairness between TCP connections and UDP connections. Besides, the setting of buffer size strongly affects the efficiency of CSFQ. In this paper, we present a novel CSFQ-based queuing algorithm. Based on the estimated value of fair share, the probability to drop packet is determined to ensure fair share of bandwidth. The simulated experimental results demonstrate that the proposed scheme outperforms CSFQ.

Improving Fairness in DiffServ Networks Using Adaptive Aggregate Markers

Traffic markers differentiate among packets from senders based on their service profile in the differentiated service networks. Researchers have previously revealed that the existing marking mechanism causes the unfairness in aggregates. This study presents a new marking algorithm. Simulation results demonstrate that the fairness of the proposed scheme exceeds that of SRTCM, TRTCM, TSWTCM and ITSWTCM for medium to high network provision levels.

An efficient architecture for Bandwidth Brokers in DiffServ networks

AbstractIn this article we examine the architecture of an entity used for automatic management and provisioning of resources for DiffServ networks. We examine the existing literature and implementations in this area, focusing on the design choices made, and we propose an architecture for the design of Bandwidth Brokers that combines an adaptive admission control algorithm for increased utilization of network resources and a mechanism for reducing the complexity overhead that intends to be both simple and effective. Specifically, we present a novel architecture for the admission control module that aims at achieving a satisfactory balance between maximizing the resource utilization for the network provider and minimizing the overhead of the module. We complement our theoretical discussion with extensive experimental simulations for the proposed Bandwidth Broker components and analysis of the results. The simulations study the possible configurations of the proposed algorithm and also compare it with alternative admission control policies. Copyright © 2007 John Wiley & Sons, Ltd.

Feedback-Based Adaptive Packet Marking for Proportional Bandwidth Allocation

Differentiated service (DiffServ) networks have been proposed to assure the achievable minimum bandwidth to aggregate flows. However, analyses in the literature show that the current DiffServ networks are biased in favor of ah aggregate flow that has a smaller committed information rate (CIR) when aggregate flows with different CIRs share a bottleneck link. In order to mitigate this unfairness problem, we propose an adaptive marking scheme which provides the relative bandwidth assurance in proportion to the CIRs of the aggregates. By introducing a virtual target rate (VTR) and adjusting it depending on the provision level of the network, each aggregate can obtain its fair share of the bandwidth, regardless of traffic load. This scheme is based on a feedback approach. It utilizes only two-bit feedback information conveyed in the packet header and can be implemented in a distributed manner. Furthermore, the proposed scheme does not require calculating fair shares of aggregates or any additional signaling protocol. Using steady state analysis and extensive simulations, we show that the scheme can provide aggregate flows with their fair shares of bandwidth, which is proportional to the CIRs, under various network conditions

An integrated end-to-end QoS anycast routing on DiffServ networks

Anycast is the point-to-point communication between a single client and the nearest destination server identified by an anycast address in traditional best-effort networks such as the Internet. Conversely, an anycast service in a QoS-based network is required to choose a path with bandwidth guarantees and to select a server with capacity guarantees. This paper proposes an approach which integrates both a network layer anycasting service and an application layer anycasting service to select a path and server that satisfies the client QoS constraint in DiffServ networks. In order to perform a path selection algorithm, the proposed protocol uses the idea of residual capacity to select all paths which match the client bandwidth requirement. In order to perform the server selection process, each server resource management (SRM) uses the server weight of each server to select the best server that has sufficient capacity for the client request. Furthermore, the design of multiple SRMs eliminates the single-point-of-failure problem and increases reliability in a DiffServ network. The simulation results have shown that the proposed protocol has much faster server response times and has the lowest request blocking rate compared to other existing protocols, especially in the case where the traffic load is high.

A framework for admission control and path allocation in DiffServ networks

We consider a Differentiated Service Domain, in which the domain administrator has to decide if to accept or to reject Bandwidth Reservation Requests (BRRs) requested by users. We first define an analytical approach and a methodology to determine the set of SLAs that can be effectively supported by a DiffServ IP network. We consider the Assured Forwarding Per Hop Behavior, and, based on the BRR probabilistic description, we derive a worst-case mathematical formulation for the overbooking probability, i.e., the probability that the traffic crossing any link of a source-destination path exceeds the link capacity. Next, we focus our attention to the problem of routing traffic arising from BRRs, i.e., the selection of paths along which traffic may flow. In particular, we show that the construction of an optimal set of paths is equivalent to the construction of a multicast tree, or a Steiner Tree, which is know to be an NP-hard problem. We therefore propose a class of simple heuristics, whose performance are assessed by simulations. Results show the effectiveness of the admission control criterium proposed, and that it is possible to increase up to 40% the amount of capacity a network provider can reserve to BRRs without violating the QoS constraints or to reduce the BRR blocking probability by a order of magnitude by using the proposed optimization algorithm.

Fair early drop marker for improving TCP fairness in multiple domain DiffServ networks

The differentiated services (DiffServ) model, proposed as a scalable way of providing quality of service in the Internet, provides packet level service differentiation on a per-hop basis. To provide the end-to-end service differentiation more efficiently for the assured services, as well as a source marker, an interdomain marker placed at the boundary edge routers connecting network domains is necessary because an Internet connection usually spans through a path involving one or more network domains. Such an interdomain marker should be operated at the aggregate flow level due to the scalability problem. The random early demotion and promotion (REDP) marker was proposed as an interdomain marker. It introduces randomness and early decisions on the three-color packet marking process at the aggregate flow level to remove the phase effect that brings about the unfairness in the demotion and promotion among different flows. The REDP marker achieves good UDP fairness in demoting and promoting packets through random and early marking decisions on packets. However, TCP fairness of the REDP marker is not obvious as for UDP sources. In this paper, we propose a new interdomain marker to resolve the problem of the REDP marker. Instead of unfair packet droppings at the RIO buffers in the core routers which cause unfairness of TCP flows, it fairly drops the green packets of each TCP flow with a token filling rate configuration method before they enter into the core routers. By doing this fair early packet dropping, it increases TCP fairness at a minor expense for the total throughput.

A fair bandwidth distribution mechanism in a Diffserv network

Previous works for the Assured Forwarding (AF) service in a Diffserv network have no sufficient consideration on the target rate and the effect of RTT and UDP in achieving the fairness of bandwidth share. Also previous works act like Best-effort service in an Under-Provisioned Network (UPN) condition. In this paper, in order to solve these problems, we propose the Proportionally Fair Differentiated Service Architecture (PFDSA), which is composed of tcp-microflow based Target rate and RTT Aware Three Color Marker (tmTRA3CM), udp-microflow based Three Color Marker (um3CM), Target Rate Based Dropper (TRBD), and Target Rate Adjusting Function (TRAF). In the results of comparisons on the performance among existing mechanisms and the PFDSA in the OPN (Over-Provisioned Network) and the UPN conditions, the PFDSA was able to further mitigate the RTT and UDP effects than the previous works. The PFDSA was shown to provide better fair bandwidth distribution proportional to various target rates even in the UPN condition.

A routing framework for load balancing of bandwidth sensitive traffic in differentiated service networks

The Differentiated Service (DiffServ) network model has been defined as a scalable framework for providing Quality of Service to applications. In this model, traffic is classified into several service classes with different priorities inside queues of IP routers. The premium service class has the highest priority. Due to the high priority of premium traffic, the global network behaviour against this service class, including routing and scheduling of premium packets, may impose significant influences on traffic of other classes. These negative influences, which could degrade the performance of low-priority classes with respect to some important metrics such as the packet loss probability and the packet delay, are often called the inter-class effects. To reduce the inter-class effects, the premium-class routing algorithm must be carefully selected such that (1) it works correctly (i.e., without loop) under the hop-by-hop routing paradigm; and (2) the congestion resulted from the traffic of premium class over the network becomes minimum. In this paper, we first introduce a novel routing framework, named compatible routing, that guarantees loop-freedom in the context of hop-by-hop routing model. Then, upon this framework, we propose two multipath architectures for load balancing of high-priority traffic on DiffServ networks. Our extensive simulations clearly demonstrate that the proposed methods distribute the premium bandwidth requirements more efficiently over the whole network and perform better than the existing algorithms, especially in the case of complex and highly loaded networks.

Measurement-Based Predictive Admission Control for Mobile QoS in DiffServ Networks

The crucial issue for mobile Internet services is to provide the guaranteed mobile QoS for multicalls. In mobile networks, we should not only guarantee handoff dropping probability but also maximize resource utilization. In this paper, we propose a QoS admission control mechanism for macro Mobile QoS based on a dynamic resource reservation. We introduce the requirements for macro Mobile QoS as mobility features and predict required resources in each routing paths thorough the combined analysis of average movement pattern and pre-defined handoff dropping probability. Our simulation results show that the proposed scheme can guarantee the pre-defined handoff dropping probability and maintain lower call blocking probability and higher resource utilization than other schemes. The propose scheme gives a practical solution for mobile networks providing mobile Internet services.

QoS Support in ARINC 664 P8 Data Networks: ATN Applications Over TCP/IP Ground-to-Ground Subnetworks

Aeronautical Telecommunication Network (ATN) has been developed by International Civil Aviation Organization (ICAO) to integrate air-ground and ground-ground data communications for aeronautical applications into a single network serving air traffic control and aeronautical operational communications. The current ICAO ATN model specifies a standalone network with its own protocols, which requires building an expensive separate network for air-ground and ground-ground communications. However, the cost of building ATN can be reduced if it can use existing networks wherever possible, such as the Internet.ARINC 664 Part 8 committee has been developing an approach for inter-operation of ATN applications over the commercial TCP/IP networks. To carry time critical information required for aeronautical applications,ATN provides different Quality of Services (QoS) to applications. Although the current Internet does not provide QoS, Internet Engineering Task Force (IETF) developed the Differentiated Services (DiffServ) network to provide differential QoS to users of next generation data networks. The objective of this paper is to investigate the possibility and conduct a proof-of-concept evaluation of providing QoS to ATN applications over the DiffServ network in the next generation Internet. Our results show that the QoS requirements ofATN applications for ground-to-ground communications can be successfully satisfied when ATN is run over a DiffServ backbone.

Per-domain packet scale rate guarantee for expedited forwarding

The expedited forwarding per-hop behavior (EF PHB) was recently replaced by a new definition, called packet scale rate guarantee (PSRG), under the Differentiated Services (DiffServ) framework. This replacement raises two challenges. One is the implementation of a PSRG server. Another is the provision of per-domain PSRG. Specifically, for the former, an open issue is whether hierarchical schedulers can provide PSRG; for the latter, it is not clear whether and how per-domain PSRG can be provided in the presence of flow aggregation. Since, in DiffServ networks, flow aggregation is a natural phenomenon and hierarchical scheduling is high-likely desired, these two challenges become even more critical. To address the first challenge, we introduce a new concept called latency-rate worst-case service guarantee (LR-WSG). We prove that, if a server provides LR-WSG, it also provides PSRG. We show that many well-known schedulers support LR-WSG, which include not only one-level schedulers but also their hierarchical versions. To address the second challenge, we first prove that PSRG can be extended from per-node to per-domain if no flow aggregation is performed. The proof is notable in that it depends solely on the concept of PSRG itself. We then investigate the provision of per-domain PSRG in presence of flow aggregation. We propose to use packet scale fair aggregator (PSFA) to aggregate flows. We show that, with PSFA, per-domain PSRG can be provided in spite of flow aggregation. We finally provide a brief discuss on the viability of using PSFA in DiffServ networks and define an expedited forwarding per-domain behavior (EF PDB).

A Premium Traffic Management in DiffServ Through End-to-End Delay Budget Partitioning

Our novel Call Admission Control (CAC) scheme [1], [2] determines the number of Expedited Forwarding (EF) (Premium) flows accepted, based on e2e available bandwidth measurement. However, it does not control the number of flows that could be accepted at a node or over a link along a selected routing path for an EF flow. Hence we complemented the functionality of CAC withoriginal traffic management algorithms [2], [3] to balance EF traffic along the selected routing path for new EF flows and consequently across Differentiated Services (DiffServ) network. The novelty of these algorithms emerges from the fact that they employ an optimization dimension of EF traffic, namely the partitioning of End-to-End (e2e) delay required for a new flow into per-link or per-DiffServ node along the selected routing path. Furthermore, some network resource policies are incorporated with these algorithms to apportion the e2e resource required for EF traffic flows intoper-DiffServ node requirement along the selected routing path. We demonstrate through analysis and simulation that our proposed optimal e2e delay budget management algorithms are superior to Equi-Partitioning (EP) algorithm for DiffServ in guaranteeing e2e delay bounds required forEF flows. Furthermore, highly loaded DiffServ nodes are assigned low delay budget and the opposite for low loaded DiffServ nodes, which result in increasing DiffServ network resource utilization.