Quality of Service in Multiservice IP Networks

While the IntServ solution to Internet QoS can achieve a strong service model that guarantees flow throughputs and loss rates, it places excessive burdens on high-speed core routers to signal, schedule, and manage state for individual flows. Alternatively, the DiffServ solution achieves scalability via aggregate control, yet cannot ensure a particular QoS to individual flows. To simultaneously achieve scalability and a strong service model, we have designed and implemented a novel architecture and admission control algorithm termed Egress Admission Control. In our approach, the available service on a network path is passively monitored, and admission control is performed only at egress nodes, incorporating the effects of cross trafic with implicit measurements rather than with explicit signaling. In this paper, we describe our implementation of the scheme on a network of prototype routers enhanced with ingress-egress path monitoring and edge admission control. We report the results of testbed experiments and demonstrate the feasibility of an edge-based architecture for providing IntServ-like services in a scalable way.

A parsimonious traffic characterisation allows the design of efficient measurement based Connection Admission Control (CAC) algorithms. In recent years the notion of effective bandwidth (EB) has been successfully employed to quantify the amount of bandwidth to allot a connection in order to meet its Quality of Service (QoS) requirements. The EBf unction depends on two parameters, namely the time scale and the space scale, whose values represent the link operating point and they are related to the link capacity, the buffer size and the traffic mix. In this paper we present a study of the Many Sources Asymptotic (MSA), a Large Deviations technique that employs the notion of EB to evaluate the performance of a queueing system. Since the MSA requires to determine the time and space parameters, we firstly analysed their sensitivity to the variation of the traffic mix. We subsequently applied the results of this analysis to the refinement of a CAC algorithm based on the MSA, by using suitable thresholds for bounding the smallest mix variation beyond which a new estimation of the link operating point is required. Finally, we compared the performance of a CAC algorithm employing first the MSA then the Large Buffer Asymptotic (LBA) to estimate the bandwidth requirement of the active calls.

In the context of an IP-over-ATM access and transport network, carrying guaranteed quality (CBR, rt-VBR) services as well as IP datagrams (as ABR or UBR traffic classes), we consider the joint problems of Call Admission Control (CAC), bandwidth allocation and routing. The presence of distributed access multiplexers is assumed, which are both geographically dispersed (e.g., at the user premises) and hierarchically structured. Such multiplexers are intelligent devices with decision making capabilities that operate jointly, in order to make the best possible use of the transport capacity of the access network and to maintain the Quality of Service (QoS) requirements of different users and service classes. Following the physical system organization, a hierarchical control structure is defined, where the admission of calls for real-time traffic classes (or different users) is performed by independent controllers; the latter are parametrized by the bandwidths allocated by a common link agent, playing the role of a “link coordinator” in the hierarchical control scheme. This decision maker aims at minimizing a general cost that captures QoS requirements both at the call-level (call blocking probability) for QoS-aware, connection-oriented services and at the cell-level (cell loss probability) for connectionless, best-effort, ones. The control architecture also reflects the multilayer hierarchy introduced by the presence of multiple teletraffic time scales, by essentially decoupling the above problem from that of ensuring QoS at the cell-level for services of the first type. We derive the optimal parameters’ setting from the numerical solution of a mathematical programming problem. Then, the same structure is applied to link multiplexers of the transport network nodes, which are supposed to possess both ATM and IP switching/routing capabilities. Routing strategies at both ATM and IP levels are defined, which are combined with the above described CAC and bandwidth allocation scheme. The performance of the whole structure is tested by simulation.

In the context of Diffserv networks some services should be characterized by end-to-end quantitative QoS guarantees. In order to provide such guarantees to single flows (or flow aggregates), the end-to-end analysis of delay and loss performance in a Diffserv domain is needed. The impact of jitter should be considered in the performance analysis at the successive nodes along the path of a flow (flow aggregate). Worst-case analysis is a solution to provide deterministic quantitative guarantees, at the price of very low efficiency. As an alternative this paper proposes a probabilistic approach, aimed at providing statistical quantitative guarantees and achieve higher efficiency. The proposed analytical approach is based on the insertion of a discarding device before the FIFO queue, called “dropper”. The purpose of the dropper is to avoid the analysis of the congestion at the burst level in the queue, allowing for the application of an analytical result derived for packet scale conflicts in the modulated ND/D/1 queue. Simulations are presented that validate the analytical bound. Finally numerical results are provided to evaluate the efficiency of the bound in an admission control scheme.

The main concern with the IETF proposed Intserv architecture is that the use of per-flow routing and reservation state in the routers may not be scalable to high-speed backbone networks. This paper proposes the Scalable Multipath Aggregated RouTing (SMART) architecture that aggregates flows along multipaths such that the per-flow reservation state in the routers is reduced to a small scalable aggregated state whose size is dependent only on the number of destinations and flow classes. The SMART architecture can be implemented in current IP networks and the complexity is similar to the best-effort IP architecture.

We introduce a control architecture in which several (independent) multimedia clusters share the same (local or metropolitan) networking resources in a controlled framework. In particular, a central entity (i.e., the Gatekeeper) harmonizes the transmission rates of the various clusters following a given sharing policy. Each cluster, in turn, adopts its own end-to-end rate control mechanism to meet the Gatekeeper’s transmission rate indications. A testbed has been developed and the system has been evaluated with real experiments by using different types of transmission sources. A software architecture is also introduced and described, with particular reference to the middleware framework realized with the Jini system.

Multimedia and real-time applications have peculiar requirements in terms of the quality of data transmission services, that are not satisfied by the best effort nature of IP. Moreover, many of the multimedia applications are characterized by a multicast communication pattern. In this work, we propose a functional architecture aiming at providing a common framework to deploy network protocols supporting QoS. The architecture is compliant with the main standards proposed in the literature. We discuss two possible implementations of the architecture modules for the service set-up, in either the int-serv model or the diff-serv model. We present some performance evaluations obtained by performing experiments with those two implementations.

In this paper we propose a DiffServa rchitecture for the support of real time traffic (e.g., video) with QoS constraints (e.g., bandwidth and delay) over an IP domain. The main goal of the paper is to identify solutions which provide QoS guarantees without requiring per flow processing in the core routers (as is commonly done in IntServs olutions) and which are thus scalable. We propose, and evaluate through simulation, different approaches for call admission control (CAC) and resource allocation. These approaches are all consistent with the DiffServmo del, but place different processing and signaling loads on edge and core routers. Paths are computed by means of a QoS routing algorithm, Q-OSPF, and MPLS is used to handle explicit routing and class separation.

In this paper, we outline a methodology that can be applied to model the behavior of TCP Reno flows. The proposed methodology stems from a Markovian model of a single TCP source, and eventually considers the superposition and interaction of several such sources using standard queueing analysis techniques. Our approach allows the evaluation of such performance indices as throughput, queueing delay and segment loss of TCP flows. The results obtained through our model are validated by means of simulation, under different traffic settings.

Guaranteeing of quality-of-service (QoS) is a challenging task to promote the evolution of the Internet from a simple data network into a true multiservice network. To this end, the IETF intserv Working Group, with the goal of defining a next generation Internet, has defined two QoS classes: Guaranteed Services and Controlled-Load Services. For both of them the source is required to declare its traffic characteristics by means of a number of Tspec parameters and guarantee these traffic characteristics during transmission. The target of this paper is to develop an analytical tool for the design of feedback laws which allow MPEG encoders to inject into the network video traffic shaped according to the declared Tspec, while maintaining an acceptable perceived quality.

This paper presents new speech coding issues and algorithms involved in IP telephony. To develop an efficient system of adaptive voice over IP (VoIP), in fact, besides traditional networking aspects, a series of speech processing issues need to be carefully considered. More specifically, two important aspects of discontinuous transmission are dealt with: the impact of the VAD on source throughput and the need for an efficient system of comfort noise. In addition we propose a variable rate, toll quality CS-ACELP coder that uses coding modes compatible with the three 6.4, 8, and 11.8 kbit/s coding schemes standardised by ITU-T in G.729. In particular, the algorithm presents 4 coding categories, with an average bit rate ranging between about 3 and 8 kbit/s, that adapt the rate to changes in network conditions.

CSELT is evaluating the feasibility of provisioning Telephony over IP (ToIP) services. ToIP is different from Voice over IP, because it is intended as a Carrier-class service, fully featured and with quality of service guarantees, to be offered to a large number of users, either overlapping or substituting classical Plain Old Telephone Services. CSELT is performing both theoretical and experimental analysis; in particular we are focusing our work on the mechanisms and protocols developed in the standardisation bodies in order to verify how the telephony services can be supported by means of an IP network. In the first half of 2000 we have performed a laboratory trial in order to demonstrate the “proof-of-concept” of ToIP capabilities, performance and interoperability with TDM-based switching technology. In the solution tested in CSELT, provided by a leading manufacturer, the IP technology is deployed in the transit network, so it is necessary to have an interworking function between a traditional circuit network and the IP network. It must be underlined that the solution provided by the manufacturer, as requested by CSELT, was still at an early stage of development, hence it has not been possible to perform a full test campaign. From our theoretical analysis and laboratory experiment we observed the suitability of architectures proposed in the standardisation bodies and by the manufacturers for the provisioning of ToIP services. The experimented system, in particular considering its prototype nature, has shown high performance in terms both of voice and signalling quality. The main problem that arises in ToIP systems is the reliability of the system, that is a fundamental concept for providing carrier-class services. The detailed results of both the theoretical analysis and the laboratory experiment are presented in this paper.

Because of their minimal complexity,Weighted Round Robin (WRR) schedulers have become a popular solution for providing bandwidth guarantees to IP flows in emerging networks that support differentiated services. The introduction of applications that require flexible bandwidth management puts emphasis on hierarchical scheduling structures, where bandwidth can be allocated not onlyto individual flows, but also to aggregations of those flows. With existing WRR schedulers, the superimposition of a hierarchical structure compromises the simplicity of the basic scheduler. Another undesirable characteristic of existing WRR schedulers is their burstiness in distributing service to the flows. In this paper, we present two enhancements for WRR schedulers which solve these problems. In the first enhancement, we superimpose a hierarchical structure bys implyre defining the way the WRR scheduler computes the timestamps of the flows. This “soft” hierarchyh as negligible complexity, since it does not require any additional scheduling layer, yet is highly effiective. The second enhancement defines an implementation of a WRR scheduler that substantiallyre duces the service burstiness with marginal additional complexity.

Input queued and combined input/output queued switchingarc hitectures must be controlled by a schedulingalg orithm, which solves contention in the transfer of data units to switch outputs. We consider the case of packet switches (or routers), i.e., devices operatingon variable-size data units at their interfaces, assuming that they internally operate on fixed-size data units, and we propose novel extensions of known schedulinga lgorithms for input queued and combined input/output queued architectures. We show by simulation that such architectures can provide performance advantages over traditional output queued architectures.

In this paper two packet discard schemes are proposed with the goal to provide incoming data flows different levels of quality of service. In particular, the assured forwarding in the Differentiated Services per-hop-behavior, as reported in the Request For Comments 2597, is considered as reference scenario.

The different levels of quality of service imply different portions of bandwidth assigned to data flows which share that same output link. Good performance in terms of throughput and fairness in bandwidth sharing among equal priority flows are also considered as the requirements to meet.

The proposed schemes are also evaluated when end-to-end transport protocols, such as TCP, are employed so as to give some informations about their impact on applications.

These dropping schemes are shown to provide good results and to represent a relatively low complex solution for dealing with the most demanding data flows during congestion intervals. Numerical results are reported and discussed for homogeneous and non homogeneous data sources.

The statistical multiplexing of non-fixed-size packet flows with heterogeneous requirements onto a single network interface of a router gave rise to a number of different scheduling mechanisms. These algorithms attempts to work as close as possible to the ideal “fluid model”. One of the most effective proposals is the Worst-case Fair Weighted Fair Queueing (WF ² Q); this pays its optimality with a great computational complexity and has been followed by more “operative” derivatives: the WF²Q+ and S-SPFQ. The former has not been specified with a univocal algorithm, thus leaving space for a number of implementations: this article aims to analyse and compare the different algorithms coming out from the WF²Q concept at both theoretical and “practical” level (by means of simulations). To this purpose, an on-purpose discrete-event simulator for packet schedulers has been implemented at the University of Pisa. The simulations also allowed to verify to what extent the operative versions of WF²Q fulfill to the WF²Q theoretical properties.

A strong trend in the Internet is to provide wireless Internet access to handheld devices. Radio spectrum is a scarce and expensive resource, and a wide-area cellular system that provides satisfying quality for a data service based on the traditional TCP/IP protocol stack will be expensive. This becomes especially problematic for real-time communication based on UDP.

To design an economically viable system, dierent approaches are available. New protocols can be designed that take wireless systems into account. The Wireless Application Protocol (WAP) is an example of such a protocol. Alternatively, the existing TCP/IP protocols can be modified to better support cellular systems. The development of WAP is tempting since it does not have to modify the TCP/IP Protocol stack. However, to provide future proong, it is a better idea to base the protocol architecture for wireless on TCP/IP since this will be the dominant Internet protocol stack for years to come.

This paper discusses requirements on various layers in the TCP/IP protocol stack to better support real-time data services. One key component is a transport layer designed to use limited bandwidth more efficiently for real-time services. Changes are required to other layers as well, but these can be deployed where needed without disturbing the rest of the Internet. By applying these changes, it becomes possible to provide cost efficient, real-time services based on IP in cellular systems.

A solution to reduce the cost of multicast paths is presented, where the traffic is split over multiple paths to obtain an overall lower cost. Optimal conditions for this split operation are provided and two different algorithms are presented. The methods were shown to work on problems with many different types of simultaneous multicast traffic. Various experiments were conducted and the results showed that the new multicasting was fairly effective on the end-to-end quality of services.

Current optical networks typically offer two degrees of service reliability: full protection in presence of a single fault in the network, and no protection at all. This situation reflects the historical duality that has its roots in the once divided telephone and data environment. The circuit oriented service required protection, i.e., provisioning of readily available spare resources to replace working resources in case of a fault. The datagram oriented service relied upon restoration, i.e., dynamic search for and reallocation of affected resources via such actions as routing table updates.

The current development trend, however, is gradually driving the design of networks towards a unified solution that will jointly support traditional voice and data services as well as a variety of novel multimedia applications. The growing importance of concepts, such Quality of Service (QoS) and Differentiated Services that provide varying levels of service performance in the same network evidences this trend.

Consistently with this pattern, the novel concept of Differentiated Reliability (DiR) is formally introduced in the paper and applied to provide multiple reliability degrees (classes) in the same network layer using a common protection mechanism, i.e., path switching. According to the DiR concept, each connection in the layer under consideration is guaranteed a minimum reliability degree, defined as the Maximum Failure Probability allowed for that connection. The reliability degree chosen for a given connection is thus determined by the application requirements, and not by the actual network topology, design constraints, robustness of the network components, and span of the connection.

An efficient algorithm is proposed to design the Wavelength Division Multiplexing (WDM) layer of a DiR ring.

This paper addresses the problem of building optical packet switches able to effectively cope with variable length packet traffic, such as IP traffic. Aswit ching architecture equipped with a multistage fiber delay lines based buffer is presented. The aim is to realize a buffer with fine granularity and long delay with an architecture of limited complexity. QoS mechanisms are proposed to support service differentiation by exploiting the wavelength resource and are evaluated by simulation.

The increasing growth and complexity of current corporate networks calls for higher level models and tools for managing network resources. In this sense, Service Level Management represents an important recent shift in system management. Furthermore, recently the IETF (Internet Engineering Task Force) DiffServ working group, has defined an architecture for implementing scalable service differentiation in the Internet, where network resources are allocated to traffic streams by service provisioning policies. This work presents and evaluates a functional architecture and a framework for mapping service management policies and constraints into DiffServ mechanisms. Various simulation scenarios described through the use of service policies are introduced and analyzed. It is shown that the use of service level management allows for efficient dynamic traffic engineering of DiffServ backbones.

Internet application performance over wireless links is disappointing, due to wireless impairments and their adverse interactions with higher protocol layers. In order to effectively address these problems without the need for global protocol upgrades, we focus on link layer enhancement schemes. Simulations reveal that different schemes work best for different applications. We have thus developed a multi-service link layer architecture that can simultaneously enhance the performance of diverse applications by supporting multiple link mechanisms concurrently. Simulations confirm that this architecture provides dramatic performance improvements. The architecture can be embedded in various ways into the Internet. A critical issue for Quality of Service support over wireless links is the unpredictability of available resources. Our approach is based on fair sharing of the link before any measures are taken to improve the performance of individual traffc classes. This approach turns over the error control trade-off to the applications themselves.

The General Packet Radio Service (GPRS) is the current enhancement in the GSM infrastructure, capable of handling Internet Protocol (IP) traffic for mobile computing and communications. A major deficiency of the current GPRS specification is the lack of adequate IP Quality of Service (QoS) support. In this paper a QoS aware architecture for enhancing the GPRS core network with IP QoS support is presented. The proposed scheme is based on the Integrated Services (Intserv) architecture, processed in the IETF during the past years. The applicability of the proposed architecture as well as the required signalling enhancements and modifications in the components of the GPRS architecture are discussed. To identify the applicability of our proposal in a real GPRS network, a simulation model of the proposed IntServ architecture was developed in order to quantify the effect that signalling overhead has on the GPRS operation and performance. The obtained simulation results show that the proposed scheme demonstrated good scalability, even for large user populations.

In a CDMA network, resource allocation is critical in order to provide suitable signal quality for each user and achieve channel efficiency. The third-generation mobile communication systems (ITU/IMT-2000) must be designed to support wideband services at bit rates as high as 2 Mbps, with the same quality as fixed networks. Mobiles transmitted power has to be controlled to provide each user a reasonable connection while limiting the interference seen by other users. Transmitted rate has also to be controlled to avoid congestion. An adaptive protocol is proposed for controlling mobile calls transmitter power and rate cooperatively when previous work has focused on handling them separately. The active component of this scheme is called Genetic Algorithm for Mobiles Equilibrium (GAME). Based on an evolutionary computational model, the base station tries to achieve an adequate equilibrium between its users. Thereof, each mobile can send its traffic with a suitable power to support it over the different path losses and interference. In the mean time, its battery life is being preserved while limiting the interference seen by neighbors. A significant enhancement in signal quality and power level has been noticed through several experiments.

Distributed end-to-end measurement based connection admission control mechanisms have been recently proposed. The goal of these schemes is to provide tight QoS control on a per connection basis by means of measurements taken by the edge nodes and priority based forwarding procedure at internal nodes. Since the additional flows handling procedures are implemented at the border routers and the forwarding mechanisms are for flows aggregates only, the approach is fully scalable and compatible with the IETF Differentiated Service proposal. The aim of this paper is to propose specific schemes and to investigate the advantages and limits of the approach by analyzing the basic mechanisms and evaluating its performance. As a results, the paper shows that end-to-end measurements taken over a low priority probing packet stream is an effective and robust way to guarantee bandwidth and delay for real-time services characterized by fast traffic dynamics, such as Voice over IP.

The resource sharing techniques in Multimedia on demand systems allow the simultaneous service of a large number of requests with considerable savings in terms of network bandwidth and server resources. In this paper, we report the results of a study that analyzes several key aspects of video distribution systems, in the hypothesis of exploiting patching techniques. Synthetically, such a technique consists in serving a client request by an existing stream for that video object (if there is any) which is buffered in the client equipment and simultaneously a new stream is requested to the server for the frames already transmitted. For such systems, a performance analysis has been carried out, exploiting analytical and simulation models. The results show the main system performance (efficiency, aggregate bandwidth, etc.) allowing to decide about the acceptance of a new service request based on QoS criteria.

Looking back at many proposals appeared on the scene in these years, a fundamental lesson to be learned is that their success or failure is strictly tied to their backward compatibility with existing infrastructures. In this paper, we consider the problem of providing explicit admission control decisions for QoS aware services. We rely the decision to admit a new flow upon the successful and timely delivery, through the Internet, of probe packets independently generated by the end points. Our solution, called GRIP (Gauge&Gate Realistic Internet Protocol), is fully distributed and scalable, as admission control decisions are taken at the edge network nodes, and no coordination between routers, which are stateless and remain oblivious to individual flows, is required. The performance of GRIP are related to the capability of routers to locally take decisions about the degree of congestion in the network, and suitably block probe packets when congestion conditions are expected. The key message of this paper is that GRIP is a novel reservation paradigm which can be seamlessly applied to the existing Diffserv (and even legacy) Internet, although a marginal increase in QoS is envisioned in these existing scenarios. Indeed, GRIP opens up a future smooth migration path toward gradually improved QoS, as routers in different domain will be upgraded with better measurement-based admission decision criteria. The enabling factor is that router decision criteria are localized and do not involve any coordination. This guarantees that they can be enhanced without losing inter-operability with installed devices.

Erratum to: M. Ajmone Marsan, A. Bianco (Eds.) Quality of Service in Multiservice IP Networks DOI: 10.​1007/​978-3-540-44554-8

The book was inadvertently published with an incorrect name of the copyright holder. The name of the copyright holder for this book is: © Springer-Verlag Berlin Heidelberg. The book has been updated with the changes.