ATM Network Simulation With OPNET

1. Introduction
Asynchronous Transfer Mode (ATM) is a connection-oriented packet switching technique that is universally accepted as the transfer mode of choice for Broadband Integrated Services Digital Network. This report describes key features of the ATM network and some relative simulation work we have done by OPNET.

2. ATM Network
Since the project is emphasize in simulation work by OPNET, well just simply introduce the basic principles of ATM: ATM is considered as a specific packet oriented transfer mode based on fixed length cells. Each cell consists of a 48bytes of information field and a 5bytes of header, which is mainly used to determine the virtual channel and to perform the appropriate routing. Cell sequence integrity is preserved per virtual channel. ATM is connection-oriented. The header values are assigned to each section of a connection for the complete duration of the connection. Signaling and user information are carried on separate virtual channels. The information field of ATM cells is carried transparently through the network. No processing like error control is performed on it inside the network. All services (voice, video, data, ) can be transported via ATM, including connectionless services. To accommodate various services an adaptation function is provided to fit information of all services into ATM cells and to provide service specific functions (e.g. clock recovery, cell loss recovery, ...). Performance Issues are the important issues we care about, There are 5 parameters that characterize the performance of ATM systems. They are 1)Throughput 2)Connection Blocking Probability 3)Cell Loss Probability 4)Switching Delay 5)Jitter on the delay Throughput: This can be defined as the rate at which the cells depart the switch measured in the number of cell departures per unit time. It mainly depends on the technology and dimensioning of the ATM switch. By choosing a proper topology of the switch, the throughput can be increased.

Connection Blocking probability: Since ATM is connection oriented, there will be a logical connection between the logical inlet and outlet during the connection set up phase. Now the connection blocking probability is defined as the probability that there are not enough resources between inlet and outlet of the switch to assure the quality of all existing as well as new connection. Cell Loss Probability: In ATM switches, when more cells than a queue in the switch can handle will compete for this queue, cells will be lost. This cell loss probability has to be kept within limits to ensure high reliability of the switch. In Internally Non-Blocking switches, cells can only be lost at their inlets/outlets. There is also possibility that ATM cells may be internally misrouted and they reach erroneously on another logical channel. This is called Cell Insertion Probability. Switching Delay: This is the time to switch an ATM cell through the switch. The typical values of switching delay range between 10 and 1000Micro Secs. This delay has two parts. 1.Fixed Switching Delay and it is because of internal cell transfer through the hardware. 2.Queueing delay and this is because of the cells queued up in the buffer of the switch to avoid the cell loss. Jitter on the Delay: This is also called as Cell Delay Variation(CDV) and this is denoted as the probability that the delay of the switch will exceed a certain value. This is called a quantile and for example, a jitter of 100Micro secs at a 10exp-9 quantile means the probability that the delay in the switch is larger than 100Micro secs. is smaller than 10exp-9. ATM Quality of Service is also we concern about, there are five ATM service classes : Constant bit rate (CBR): CBR ensures that all cells in a transmission are maintained from end to end. This service type is used for voice and video transmission that require little or no cell loss and rigorous timing controls during transmission. Real-Time Variable Bit Rate (rt-VBR): rt-VBR is intended for time-sensitive applications, (i.e., those requiring tightly constrained delay and delay variation), as would be appropriate for voice and video applications. Non-Real-Time (nrt-VBR): nrt-VBR is intended for applications which have bursty traffic characteristics and do not have tight constraints on delay and delay variation. Available bit rate (ABR): ABR supports available-bit-rate traffic, minimum guaranteed transmission rate, and peak data rates. ABR also allows bandwidth allocation depending on availability, and it uses flow control to communicate bandwidth availability to the end node.

Project Report for EE 5570: Communication Networks

Xiangqian Yu

Unspecified bit rate (UBR): UBR is the best-effort service intended for non-critical applications, which do not require tightly constrained delay and delay variation, nor a specified quality of service.

2. Simulation by OPNET
2.1 Introduction: OPNET(Optimized Network Engineering Tool) provides a comprehensive development environment for the specification, simulation and performance analysis of communication networks. OPNET provides four tools called editors to develop a representation of a system being modeled. These editors, the Network, Node, Process and Parameter Editors, are organized in a hierarchical fashion, which supports the concept of model level reuse. Models developed at one layer can be used by another model at a higher layer. 2.2 ATM Model Features: Signaling Support : Signaling is provided for point-to-point, full-duplex, Switched Virtual Circuit (SVC), Soft-Permanent Virtual Circuit (SPVC) and Soft-Permanent Virtual Path (SPVP). Dynamic call setup and teardown procedures are supported. Traffic Control: Traffic control includes Call Admission Control (CAC) and Usage Parameter Control (UPC). Traffic Control is based on specific service category, traffic parameters (PCR, SCR, MCR,MBS) and QoS parameters (ppCDV, maxCTD,CLR). Traffic Control prevents all calls with unsupportable traffic requirements from establishing a connection and prevents established calls from degrading network service below the specified Quality of Service requirements. Buffering: Output port buffering is modeled. Output buffer supports the round-robin and weighted round-robin queuing schemes. Buffers can be configured at each switch for various QoS levels. A QoS level is made up of the QoS category (CBR, rt-VBR, nrt-VBR, ABR, UBR), the QoS parameters, (ppCDV, max CTD, CLR), and the traffic parameters. 2.3 ATM Node Models: The ATM model suite contains several client and server node models, which can be subdivided into the following categories: workstations and servers, uni-clients and uni-servers, uni-sources and uni-destinations, and intermediate switching elements (such as clouds and

Project Report for EE 5570: Communication Networks

Xiangqian Yu

switches). The ATM nodes can be found in the object palettes with an atm prefix: atm, atm_advanced, atm_lane, and atm_lane_advanced. In this simulation, we choose atm_uni_client for applications that run directly over ATM. Uni-clients and uniservers: atm_uni_client, atm_uni_server These client node models feature an application layer that resides directly over the ATM layer. Unlike the ATM workstation node model, the ATM uni-client model establishes a separate ATM connection for each application task. Examples of application tasks are sending an e-mail, downloading a .le, and making a voice call. ATM uni-clients can be used only with ATM uni-servers, which are capable of supporting all of the application services. The ATM uni-client and uni-server node models are located in the atm_advanced object palette. 2.3 ATM Model attributes: The intermediate and advanced ATM nodes have several attributes that can be used to specify ATM con.guration details. Some of the important ATM model attributes we concern in our simulation are: 1. Traffic Contract: . This attribute specifies the traffic contract used by the application layer when it sends traffic over an ATM stack. Although the application layer includes data traffic, signaling traffic and IP/ATM routing traffic, only data traffic has a configurable traffic contract. The traffic contracts for signaling traffic, IP/ATM routing traffic and LANE control traffic are not configurable. The parameters specified in a contract do not account for overheads from the AAL (service dependent) and ATM layers (5 bytes). The Traffic Contract attribute has 3 parts: the Category, the Requested Traffic Contract, and the Requested QoS. Category: This attribute specifies the service category used by the application. OPNET supports all five categories specified by the ATM Forum Traffic Management Specification 4.0: CBR, rt-VBR, nrt-VBR, ABR, and UBR. For a call to be admitted by call admission control, there should be at least one path to the destination where all nodes support the requested service category. Requested Traffic Contract: This attribute specifies the traffic parameter settings for the connection. The Requested Traffic Contract allows you to specify the peak cell rate (PCR), minimum cell rate (MCR), sustainable cell rate (SCR), and mean burst duration (MBS) in the incoming and outgoing directions. During call admission control, these requested values are compared to the supported parameters on all intermediate nodes. Requested QoS: This attribute specifies the applications requested Quality of Service, which includes the peak-to-peak cell delay variation (ppCDV), the maximum cell transfer delay (maxCTD), and the cell loss ratio (CLR). During call admission control, these requested values will be compared to the supported parameters on all intermediate nodes. 2. Port Buffer Configuration:

Project Report for EE 5570: Communication Networks

Xiangqian Yu

This attribute is used to specify supported parameters and to con.gure buffers on each port of a node. The configuration specified in this attribute applies to all ports of the node. Queue Number: This attribute specifies the queue index. To automatically assign indices to the queues, you can use the Per VC setting. Alternatively, you can assign each queue a unique queue number. The queue number is used to identify the queue being monitored for certain statistics (such as queue length). Queue Parameters: This attribute allows you to specify the amount of bandwidth that is allocated to a specific queue. Max_Avail_BW (% Link BW): This is the maximum bandwidth available to this queue. It is calculated as a percentage of the link bandwidth. For CBR calls, this attribute regulates the maximum bandwidth reserved and hence guarantees this bandwidth as well. Min_Guaran_BW (% Link BW): This is the minimum guaranteed bandwidth expressed as a percentage of link bandwidth. For non-CBR calls, this attribute defines the bandwidth reserved. For example, for a rt-VBR call, SCR is the minimum guaranteed bandwidth. The value specified for minimum guaranteed bandwidth is equal to the weight of this queue when the ATM QoS Priority Scheme attribute is set to weighted round-robin. Size: This attribute determines the number of cells in the queue. Traffic and Qos Parameters: These attributes are used when selecting a buffer for a call that is waiting for admission through a port in a node. A buffer is selected if both of the following requirements are met: The traffic parameters (PCR, SCR, MCR, and MBS) of the incoming call are less than or equal to the value specified in the Traffic Parameters attribute. The QoS parameters (maxCTD, ppCDV, CLR) of the incoming request are greater than or equal to the values specified in the QoS Parameters attribute. 3. ATM Switching Speed: This attribute specifies how fast a cell is switched through the core switching fabric. This speed is specified in cells/sec. 4. ATM QoS Priority Scheme: This attribute specifies the servicing scheme for the queues. Two types of queuing schemes are available: round-robin and weighted round-robin. Weighted round-robin scheme: queues are serviced depending on the weights assigned to them. Weights are determined according to the Minimum Guaranteed Bandwidth attribute (in ATM Port Buffer Configuration >Queue Parameters attribute) of each queue parameter. This scheme ensures that the guaranteed bandwidth isreserved. Round-robin scheme: all queues have the same priority and therefore have the same chance of being serviced. The links bandwidth is equally divided amongst the queues being serviced. 5. Routing attributes:

Project Report for EE 5570: Communication Networks

Xiangqian Yu

Switches in the ATM model suite use a dynamic routing protocol, ATM Distance Vector Routing, which is implemented as a distributed, asynchronous adaptation of the Bellman-Ford shortest path algorithm. When the ATM signaling layer receives a call setup request, the source node finds a route to the calls destination. The following attributes are used to configure this routing protocol: ATM Routing Update Interval. This attribute specifies the time between regular routing updates. Routing tables are periodically updated to ensure that all nodes are aware of the latest topology changes. ATM Active and Passive Failure Detection Modes. Failures and recoveries in the network must be detected by nodes adjacent to the failure or recovery point. These (adjacent) nodes must then inform other nodes in the network of the failure or recovery. active failure detection mode: the routing process detects a neighbor node or link failure/recovery and updates its routing tables immediately. The node sends out route advertisements that reflect its updated routing table. passive failure detection mode: failure is detected implicitly when no route costs have been received in two or more route update periods. 2.4 Simulation results: The fig1 is the ATM network we constructed by OPNET. This ATM network we are simulating consists of several ATM switches, server and clients(the reason why we didnt use more ATM switches is the simulation time increase beyond our computer can afford), we use OC3 link to connect the network for supporting maximum 155Mbps traffic.

Project Report for EE 5570: Communication Networks

Xiangqian Yu

Fig1. ATM network structure

We can simply adjust the traffic load by changing the Traffic Scaling Factor in Configure Simulation menu before we run the simulation every time. The three traffic components: video, voice and data are generated, we use rt_VBR for video, CBR for the voice, and ABR for data traffic. However, the ratio of the three kinds of traffic are difficult to set to exactly 30%, 40% and 30%. Fig2 is the comparison of the traffic generated by video, voice and data.

Project Report for EE 5570: Communication Networks

Xiangqian Yu

Fig2. Traffic load comparison

The following tables are the statistic results of data, video and voice sevices. Statistic Ftp Download Response Time (sec) Ftp Traffic Received (bytes/sec) Ftp Traffic Received (packets/sec) Ftp Traffic Sent (bytes/sec) Ftp Traffic Sent (packets/sec) Ftp Upload Response Time (sec) Statistic Average 0.072 84,309 30.7 84,814 30.8 0.070 Maximum 0.109 285,440 96.7 242,267 93.3 0.107 Minimum 0.059 0 0.0 0 0.0 0.059

Average Maximum Minimum 0.0709 0.0708 0

Video Conferencing Packet End-to-End Delay 0.0708 (sec) Video Conferencing Traffic

Received 313,018 691,200

Project Report for EE 5570: Communication Networks

Xiangqian Yu

(bytes/sec) Video Conferencing (packets/sec) Traffic Received 18.1 40.0 0.0 0 0.0

Video Conferencing Traffic Sent (bytes/sec) Video Conferencing Traffic Sent (packets/sec)

315,350 691,200 18.3 40.0

Statistic Voice Packet Delay Variation Voice (sec) Packet End-to-End Delay

Average

Maximum

Minimum

0.000000007 0.000000014 0.000000004 7 6 2 0.0237 13,344 404 13,376 405 0.0237 23,100 700 23,100 700 0.0235 0 0 0 0

Voice Traffic Received (bytes/sec) Voice Traffic (packets/sec) Received

Voice Traffic Sent (bytes/sec) Voice Traffic Sent (packets/sec)

The shortest path routing algorithm we are using in OPNET is a dynamic routing protocol, ATM Distance Vector Routing. Since the OPNET only support two types of queuing schemes: round-robin and weighted round-robin, and the simulation is limited by the time constriction, we can only compare the two queuing schemes running for 30 seconds processing time(the actual time for the simulation is much longer, may take half an hour to several hours). The following two tables are the results for running at two different queuing schemes. Average ATM Call Blocking Ratio (%) 0 ATM cell Delay (sec) 0.00370 ATM throughput (bits/sec) 1,390,592 Table1. Statistic results in round-robin Statistic
ATM Call Blocking Ratio (%) ATM cell Delay (sec) ATM throughput (bits/sec)

Statistic

Maximum 0 0.00755 3,064,107

Average 0 0.00387 1,390,239

Maximum 0 0.00910 3,064,453

Project Report for EE 5570: Communication Networks

Xiangqian Yu

Table2. Statistic results in weighted round-robin

Conclusion:
The software simulation package, OPNET, which specializes in discrete-event simulation of communication systems, has many attractive features and can simulate large communication networks with detailed protocol modeling and performance analysis. In our study, the performance of different ATM switch buffer queuing schemes for roundrobin and weighted round-robin didnt have too much difference. In conclusion, discrete-event simulation provides detailed, accurate network simulation results and can observe a wide variety of network statistics. However, this method of simulation generally requires ample significant time and memory. Comments: Since the public machine in the EE department cant run the OPNET simulation properly, all the simulation work in this report is done in my own PC at home, so the simulation results may be not perfect enough, the time and the memory have a strong impact on the simulation.

Reference: [1]http://www.microsoft.com/windows2000/en/server/help/default.asp? url=/windows2000/en/server/help/sag_ATM_und_atmqos.htm [2] http://cne.gmu.edu/modules/atm/Texttut.html [3]Network simulation with OPNET: http://www.informs-cs.org/wsc99papers/043.PDF

Project Report for EE 5570: Communication Networks

Xiangqian Yu

10