2.4 Models to quantify the CDV in ATM networks 2.4.1 Single queues 2.4.2 Tandem queues 2.4.3 Impact of CDV on resource allocation 3 Cell Delay Variation introduced in a real network. EXPLOIT and BAF Testbeds. 3.1 Delay and CDV in an ATM Access Network. The BAF Testbed 30 3.1.1 ATM-PON system architecture 31 3.1.2 The MAC protocol 31 3.1.3 APON slot format 33 3.1.4 The BAF Testbed 35 3.1.5 Performance of the Access Network 36 3.2 Experimental evaluation of CDV impact on ATM resource management. The EXPLOIT Testbed 39 3.2.1 Network configuration 40 3.2.2 Traffic sources 43 3.2.3 Point Measurements on the Testbeds 46 3.2.4 Fixed delays and CDV without background traffic 48 3.2.5 CDV with background traffic 49 3.3 Conclusions and comments about the experiments 54 4 The Benes approach applied to ATM Networks. 56 4.1 The Benes approach to the virtual Waiting Time 56 4.2 Queueing models at cell level using the Benes approach 61 4.2.1 The M/D/1 system 61 4.2.2 The N • D/D/l system 4.2.3 The modulated N • D/D/l system 4.2.4 The M + D/D/l queue system 67 4.2.5 Other queue systems at cell level 4.3 Queueing models at burst level using the Benes approach 68 4.4 The ballot theorem applied to periodic queues 8 Resource management. 120 8.1 The concept of negligible CDV 8.2 Traffic shaping mechanisms 8.3 A simple model to study the spacing of WCT sources 8.4 Trade-offs in the design of a CAC combined with a Traffic-Shaping mechanism 8.5 Performance evaluation of the CAC combined with a traffic-shaper 8.6 Scheduling algorithms 8.6.1 Weighted Fair Queueing -Packet by Packet Generalized Processor Sharing 8.6.2 Virtual Spacing -Self-Clock Fair Queueing 8.6.3 Scheduling Algorithms in ATM 133 8.7 Conclusions and comments 9 General conclusions and comments. Acronyms. Bibliography. 8.1 Shaper in the multiplexor only applied to high bit rate connections 8.2 Admissible load for a buffer size of 128 and CLR=1CT 9 8.3 CAC algorithm flowchart 8.4 CAC curve for T mai =80, Buffer Size=128 cell slots, CLR=10~9 8.5 Shaped traffic versus all the traffic with the former QoS conditions 8.6 Shaped traffic versus all the traffic with T mai =150 connections 8.7 Delay on the 25 Mb/s connection 135 8.8 Delay on the 2 Mb/s connection 135 8.9 Delay on the WCT source 135 8.10 Delay on the 10 Mb/s CBR source 135 8.11 Delay on the WCT source for the VS and its modification 137 8.12 Delay on the 10 Mb/s CBR source for the VS and its modification 137