In theceZWWN of Di#erentiatedServicn the Assured Forwardingcorw recward proposed by the IETF o#ers the ceZfiN%W#ZEF of distinguishing two or morecreZAz ofpacF#AW based on bit marking inpac%fi headers.

In this paper we present a simulation study of HTTP traffic crossing a DiffServ domain. We consider both the cases where the reserved bandwidth is not exceeded by the offered traffic (overprovisioning) and where the assured traffic competes with the classic Best Effort class (underprovisioning). Simulation reported shows that DiffServ approach is able to protect the assured flows in the first case, while the performance benefits are tighter in the second case, in which fairness issues arise between long and short-lived flows.

In today's Internet, only best-effort service is provided. With up-coming Quality of Service (QoS) requirements raised by a wide range of communication-intensive, real-time multimedia applications, the best-effort service is no longer sufficient. As a result, Differentiated Service Model (DiffServ) has been proposed as a cost-effective way to provision QoS in the Internet.

Advances in IP-technology, which include the real-time encoding and decoding of voice and video, router mechanisms, and real-time transport protocol (RTP) offer great potential for computer-telephony integration over IP networks. However, due to the unpredictable loss, delay, and delay jitter in packet switching network, the conventional Internet single best-effort class of service can adversely impact the end-to-end performance of real-time applications. This dissertation proposal describes a research effort in determining mechanisms to provide Quality of Service (QoS) in an efficient and scalable manner over the Internet. We will elaborate on an end-to-end service architecture that can support per session statistical quality of service guarantees, by building on the significant amount of previous research on network- and application-level QoS solutions. In our initial work, we have chosen Voice over IP (VoIP) traffic as a workload for evaluation, and use a subjective test to determine the effects of network centric parameters on human perceived voice quality. First, we develop resource allocation techniques for VoIP traffic in virtual private networks (VPNs) to achieve the required QoS. Specifically, we will show how to use traffic statistics in predicting aggregate bandwidth usage, which is crucial for capacity planning and admission control decisions. Since these

this paper we will refer to in-profile traffic as Green, while Yellow class will be used to represent both out-of-profile and Best Effort traffic The main contribution of this paper is twofold: the proposal a Fair Marker that, by using a limited amount of memory in edge nodes, achieves the same performance enhancement of a perflow marker, but with a much simpler design; also, a simple analytical model of the completion time of short-lived flows in a DiffServ domain using our proposed Fair Marker

This dissertation presents an exploration on solving traffic load balancing problems concentrating on the IP layer in hop-by-hop networks. The first part of the traffic load balancing research is focused on bandwidth-sensitive routing for premium class traffic in Differentiated Service (DiffServ) networks, where bandwidth usage of the premium class traffic is critical not only for the traffic itself, but also for other classes of traffic with lower priorities in the same network, such as the assured or best effort traffic. If the traffic in a network is splittable, then the second part of the load balancing research is intra-domain traffic engineering in the network, where the bandwidthsensitive routing solutions in the first part can be used to achieve better traffic load balancing results in this part.

In Differentiated Services networks, packets may receive a different treatment according to their Differentiated Services Code Point (DSCP) label. As a consequence, packet marking schemes can be devised to differentiate packets belonging to a same TCP flow, with the goal of improving the experienced performance. This paper presents an analytical model for an adaptive packet marking scheme proposed in our previous work. The model combines three specific sub-models aimed at describing i) the TCP sources aggregate ii) the marker, and iii) the network status. Preliminary simulative results show quite accurate predictions for throughput and average queue occupancy. Besides, the research suggests new interesting guidelines to model queues fed by TCP traffic.

The differentiated services (DS) architecture has been proposed for providing different levels of services and has recently received wide attention. A packet in a DS domain is classified into a class of service according to its contract profile and treated differently by its class. There are currently two types of service standardized, Expedited Forwarding (EF) and Assured Forwarding (AF). In this dissertation, we focus on AF service.

Grid (GIG) transport network, packet handling must provide preferential transport to high Precedence traffic under all networking conditions, specifically conditions of resource scarcity, e.g., network overload conditions, while simultaneously satisfying packet scheduling required to meet application Quality of Service (QoS) needs. Our approach to this duality is to enhance Active Queue Management (AQM) techniques to provide Precedence and Preemption (P&P) capabilities and rely upon standard, well studied QoS Per Hop Behavior (PHB), e.g., Weighted Round Robin, Class-Based Fair Queuing, etc., for handling QoS requirements. In this way, when operating under engineered loads, the well known scheduling algorithms support high quality QoS for applications. Under network congestion situations, the enhanced AQM layer provides the necessary P&P preferential packet handling favoring high Precedence-Level (P-L) information. Our scheme allows low order queues (within the context of QoS handling) to plead up to the next higher order queue for help in alleviating queue congestion under periods of communication link overload. We refer to our scheme as the Cross Queue-AQM (CQ-ACM) Scheme. Our scheme can be extended to higher numbers of queues and any type of scheduler in a straightforward manner. Through extensive simulation studies and analytical modeling, we investigate the performance of our CQ-AQM scheme under heavy traffic limits, where Preemption is required. The performance metrics of interest to our analysis are packet delay, packet loss and throughput as a function of the packet QoS class and P&P level. Our previous studies concentrated on general nonflow controlled traffic and showed that our algorithms performed extremely well. In this paper we extend our analysis to flowcontrolled traffic by incorporating TCP traffic models into our simulation studies. We find that the application of our CQ-AQM scheme on top of standard QoS scheduling is effective in simultaneously supporting QoS and P&P transport for TCP flows as well.

Abstract—The desire or ability of an ISP to provide differentiated service is a current hotly debated topic. In this paper, we quantify the value of having differentiated service (i.e., class-of-service (CoS)) support in an IP backbone. We compare the capacity requirements of a Diffserv environment providing service for applications that require delay or loss assurances in comparison to a network that provides classless (i.e., besteffort) service and still has to meet the same performance assurances. Our modeling framework first develops a link model that quantifies the Required Extra Capacity (REC) in order for a classless link to provide the same level of performance as experienced by premium class traffic passing through a fixed capacity CoS link. We develop the REC calculations for the cases when average delay or the average loss probability is the target performance goal with Poisson or Markov Modulated Poisson Process (MMPP) input traffic. Our primary contribution is in quantifying the value of the CoS support in a network setting. I.