No. 2 (2004)
Published:
2004-06-30
ARTICLES FROM THIS ISSUE
-
Fairness considerations with algorithms for elastic traffic routing
Abstract
The bit rate of modern applications typically varies in time. We consider the traffic elastic if the rate of the sources can be controlled as a function of free resources along the route of that traffic. The objective is to route the demands optimally in sense of increasing the total network throughput while setting the rates of sources in a fair way. We propose a new fairness definition the relative fairness that handles lower and upper bounds on the traffic rate of each source and we compare it with two other known fairness definitions, namely, the max-min and the proportional rate fairness. We propose and compare different routing algorithms, all with three types of fairness definitions. The algorithms are all a tradeoff between network throughput, fairness and computational time.
-
An exact algorithm for design of content delivery networks in MPLS environment
Abstract
Content delivery network (CDN) is an efficient and inexpensive method to improve Internet service quality. In this paper we formulate an optimisation problem of replica location in a CDN using MPLS techniques. A novelty, comparing to previous work on this subject, is modelling the network flow as connection-oriented and introduction of capacity constraint on network links to the problem. Since the considered optimisation problem is NP-complete, we propose and discuss exact algorithm based on the branch-and-cut and branch-and-bound methods. We present results of numerical experiments showing comparison of branch-and-cut and branch-and-bound methods.
-
PFS scheme for forcing better service in best effort IP network
Abstract
The paper presents recent results corresponding to a new strategy for source traffic generating, named priority forcing scheme (PFS), allowing Internet users for getting better than best effort service in IP network. The concept of PFS assumes that an application, called PFS application, sends to the network a volume of additional traffic for the purpose of making the reservations for the data traffic in the overloaded router queues along the packet path in the IP network. The emitted redundant packets, named R-packets, should be rather of small size comparing to the data packets, named D-packets. The PFS scheme assumes that the R-packets waiting in a queue can be replaced by the arriving D-packets and belonging to the same flow. In this way, the D-packets can experience a prioritised service comparing to the packets produced by a non-PFS application. Notice that the proposed solution does not require any quality of service (QoS) mechanisms implemented in the network, like scheduler, dropping, marking etc., except R- and D-packets identification and replacing. We discuss the PFS efficiency for forcing priority in the overloaded conditions. Moreover simple system analysis is also presented. Finally, the profits of using PFS scheme are illustrated by examples corresponding to FTP (TCP controlled traffic) and VoIP (UDP streaming traffic) applications.
-
Methods for evaluation packet delay distribution of flows using Expedited Forwarding PHB
Abstract
The paper regards problem of providing statistical performance guarantees for real-time flows using Expedited Forwarding Per Hop Behavior (EF PHB) in IP Differentiated Services networks. Statistical approach to EF flows performance guarantees, based on calculation of probability that end-to-end packet delay is larger than certain value, allows larger network utilization than previously proposed deterministic approach. In the paper different methods of packet delay distribution evaluation are presented and compared. Considered cases comprise evaluation of delay distribution models for the core network and evaluation of end-to-end packet delay in the network consisted of edge node and chain of core nodes. Results obtained with aid of analytical models are compared with simulation results.
-
Application of a hash function to discourage MAC-layer misbehaviour in wireless LANs
Abstract
Contention-based MAC protocols for wireless ad hoc LANs rely on random deferment of packet transmissions to avoid collisions. By selfishly modifying the probabilities of deferments greedy stations can grab more bandwidth than regular stations that apply standard-prescribed probabilities. To discourage such misbehaviour we propose a protocol called RT-hash whereby the winner of a contention is determined using a public hash function of the channel feedback. RT-hash is effective in a full hearability topology, assuming that improper timing of control frames is detectable and that greedy stations do not resort to malicious actions. Simulation experiments show that RT-hash protects regular stations` bandwidth share against various sophisticated greedy strategies of deferment selection; as such it may contribute to MAC-layer network security.
-
Approximate performance analysis of slotted downlink channel in a wireless CDMA system supporting integrated voice and data services
Abstract
This paper is concerned with the performance analysis of a slotted downlink channel in a wireless CDMA communication system with integrated packet voice and data transmission. The system model consists of mobile terminals (MT) and a single base station (BS). It is assumed that the voice (data) packet error rate (PER) does not exceed 10-2 (10-5). With this requirement the number of simultaneous transmissions over the downlink channel is limited. Therefore, the objective of the call admission control is to restrict the maximum number of CDMA codes available to voice and data traffic. Packets of accepted voice calls are transmitted immediately while accepted data packets are initially buffered at the BS. This station distinguishes between silence and talkspurt periods of voice sources, so that data packets can use their own codes for transmission during silent time slots. Data packets are buffered in queues created separately for each destination. Discrete-time Markov processes are used to model the system operation. Statistical dependence between queues is the main difficulty which arises during the analysis. This dependence leads to serious computational complexity. The aim of this paper is to present an approximate analytical method based on the restricted occupancy urn model which enables to evaluate system performance despite the dependence. Numerical calculations compared with simulation results show excellent agreement for the average system throughput and the blocking probability of data packets for higher system loads. On the other hand, when the average data packet delay is considered, analytical results underestimate simulation and therefore only approximate system performance evaluation is possible.
-
Analysis of bandwidth reservation algorithms in HIPERLAN/2
Abstract
This paper focuses on performance of channel access methods in the HIPERLAN/2 standard. It discusses commonly used approaches to bandwidth allocation and presents a modified algorithm for effective bandwidth management based on pre-scheduled resource grants. Simulation results show that the new algorithm ensures much higher throughput compared to the standard method.
-
An adaptive hidden Markov model for indoor OFDM based wireless systems
Abstract
Detailed physical layer simulation of orthogonal frequency division multiplexing (OFDM) systems requires programs that execute too slowly due to long coherence time of the indoor mobile channel. Evaluation of higher layers of such systems is simplified if suitable models for reproduction of channel errors statistics are available. An adaptive hidden Markov model (HMM) for indoor OFDM based systems that accurately reproduces error statistics of the real system with less computational effort than the exact simulation is presented in this paper. The standard HMM methodology has been modified in order to reproduce the periodicity in the error positions of the OFDM systems. The proposed model is validated by comparison of three statistical parameters: number of errors, length of the errors run and length of the error-free intervals in a frame of bits.
-
Testing of interworking between network terminals with FSK receivers and public exchanges providing display and related services
Abstract
Implementation of new services in the network requires appropriate methods and tools for checking correctness of interworking between terminals and exchanges. In this article the methodology and test procedures prepared for testing terminals handling FSK protocol transmitted over the local loop for display and related services was described. Methodology presented here is based on ETSI standards. Test procedures were developed for services offered in Polish network. Tests cover all levels of FSK protocol. For two lower layers, separate procedures for ``on-hook`` and ``off-hook`` loop states were prepared. The procedure for ``on-hook`` state contains tests related to data transmission ``associated`` and ``not associated with ringing``. These procedures cover normal cases with parameter values and sequence elements complying with standards and exceptional procedures with the extreme values of the parameters and with modified elements of the sequences.
-
Labeling of signals in optical networks and its applications
Abstract
The paper is a review and comparative analysis of most common techniques proposed to attach additional data or identification information to digital signals in optical fiber networks by purely optical means. Such ``labels`` or ``headers`` can be attached either to continuous bit streams, e.g., in SDH networks or to optical packets. They enable to monitor, route and identify signals in transparent optical networks, especially those with optical wavelength multiplexing, allow management and supervision of remote optical amplifiers and can be used in optical switching systems. Other applications of this relatively unknown technology include monitoring of optical path dispersion, equalization of channels in DWDM systems and detection of intrusion or jamming in highly secure networks.
