The ATM Forum recommends the use of the Generic Cell Rate Algorithm (GCRA) to perform Usage Parameter Control at the User Network Interface of ATM networks. In order to facilitate the Call Admission Control and resource allocation procedure, it is important to investigate the characteristics of the model in which GCRA-enforced sources are merged together by a multiplexer. Such a multiplexer could be the one arranged in front of a switch to concentrate user traffic and reduce the number of required input ports. It may also represent the logical multiplexer at the output port of a switch that collects cells routed from various input ports. Moreover, it may represent the service function of the edge router situated between the integrated-services (IntServ) networks and the backbone networks that provide differentiated-services (DiffServ). In this paper, the environment under discussion is a multiplexer in which every traffic source is enforced by a dual-stage GCRA enforcer before being merged. The worst traffic pattern that maximizes the average waiting time in the multiplexer is found. The maximum average waiting time is deduced and expressed as a function of the GCRA parameters and the number of multiplexed sources. In particular, the analysis considers the speed-up function, which is widely used for ATM multiplexers and switches. The results can also be applied to a GCRA shaper without any modification.

Real-time services, including constant bit rate (CBR) and real-time variable bit rate traffic (rt-VBR), have become increasingly important owing to the rapid proliferation of multimedia applications. A cell multiplexing method capable of handling real-time traffic should satisfy related quality of service (QoS) requirements, including cell transfer delay (CTD), cell delay variation (CDV) and cell loss ratio (CLR). In this paper, we present an efficient cell multiplexing method, called longest delay beyond expectation (LDBE), to schedule real-time and non-real-time traffic in ATM networks. For the real-time traffic, LDBE scheme can minimize the CDV, and reduce the CLR and CTD, particularly when different CDV tolerance (CDVT) values are applied at each node along the path of a connection. Simulation results demonstrate that the proposed LDBE performs better than other multiplexing methods regarding these CLR, CDV and CTD criteria for real-time traffic. Furthermore, the proposed LDBE is also suitable for scheduling non-real-time traffic by providing a low CLR for non-real-time variable bit rate (nrt-VBR) and minimizing the CTD for unspecified bit rate (UBR) traffic.

We introduce the concepts of conservative cell loss ratio (CLR) estimation and worst case cell arrival patterns, and apply them to cell arrival patterns that conform to the generic cell rate algorithm (GCRA). We define new sets of cell arrival patterns which contain the worst case cell arrival patterns for conforming cell arrival patterns. Based on these sets, we propose an upper bound formula using the burst tolerance as well as peak cell rate and sustainable cell rate, and develope a connection admission control method that guarantees cell loss ratio performance satisfying its objective.

The proposed GCRA (Generic Cell Rate Algorithm) traffic shaper consists of a regulator and a scheduler. It can shape multiple incoming VBR (Variable Bit Rate) cell streams simultaneously to be strictly conforming according to the GCRA algorithm when the cells depart for the ATM output link. The impact of cell emission conflicts is considered and resolved by using an EDD (Earliest-Due-Date) scheduler and a feedback signal from the scheduler to the regulator. The call admission control condition and the cell delay bound are derived. Simulation results demonstrate that the output cell streams of the proposed GCRA traffic shaper do not contain any non-conforming cells and the scheduler queue size is significantly reduced. Meanwhile, the delay performance is almost not affected by the use of the feedback mechanism.

Due to the Cell Delay Variation (CDV) at User Network Interface (UNI), it is very hard for an ATM network to perform Usage Parameter Control (UPC), which is an important job for congestion control. Based on the Generic Cell Rate Algorithm (GCRA), ATM Forum has proposed a procedure to perform the UPC. However, the severe problem is that a user has to specify the CDV Tolerance at the UNI by itself. Such a nearly unreachable constraint makes the GCRA unsuitable for UPC. In this paper, we point out that the CDV comprises two parts in which the customer and a network provider should be responsible. Thus, we propose a concept of Innocent Public Network and an Agent Protocol to realize the principle and facilitate UPC. In addition, a shaper is suggested for the customer to employ so as to prevent its performance degradation. In the proposed system, the network is no longer suffered from CDV at the UNI and the UPC can be easily preformed.

In ITU-T Recommendation I.371, the Generic Cell Rate Algorithm (GCRA) is used to define Peak Cell Rate for the ATM network. It is further applied by the ATM Forum '93 to define Sustainable Cell Rate and Burst Tolerance so as to facilitate Usage Parameter Control and Network Parameter Control. To judge the validity of a cell according to declared GCRA parameters, the enforcer must read the clock time when the cell arrives. However, the clock of the enforcer would roll over frequently and accordingly the judgment would be incorrect. On the other hand, for a shaper in a customer premise node to dispatch cells conforming to the declared GCRA parameters, the clock would also roll over and the cell would not be dispatched correctly. To overcome the problems induced by clock roll-over, based on "time difference" concept, we propose two modified GCRA's for the enforcer and shaper, respectively. According to the proposed algorithms, we design a feasible architecture for a multi-connection shaper and simplify it for an enforcer. They are proven to perform well in spite of the inherent clock roll-over characteristics. By simulation, we evaluate the delay in the shaper and the loss in the enforcer. The features of the architectures are also discussed.