International Global Navigation Satellite Systems Society IGNSS Symposium 2011 University of New South Wales, Sydney, NSW, Australia 15 17 November 2011 Development of MF-TDMA based Satellite Network System Simulation Model Sung-Hyung Lee Department of Electronics Engineering, Ajou University, Korea +82-31-219-2474, xaviersr@ajou.ac.kr Jae-Hyun Kim Department of Electronics Engineering, Ajou University, Korea +82-31-219-2477, jkim@ajou.ac.kr ABSTRACT This paper describes about development of MF-TDMA based satellite network system simulation model. This model consists of earth station nodes accessing and controlling the network by MF-TDMA scheme. This model also includes channel model of GEO-satellite based communication system such as delay, path loss, atmosphere absorption, and ionospheric effect. Moreover, this model includes dynamic rain fading model with four state Markov chains which models the time interval, fading depth, and variation of rain fading. To verify our simulation model, we gather the changes of signal-to-noise ratio of signal receiver and compare it with rain fading dynamic of real satellite network system. We also measure the network delay using ping command, and throughput using file transmission by FTP, which can be possible by system-in-the-loop function of OPNET simulator KEYWORDS: MF-TDMA, Network Simulation Model, Satellite Network, OPNET 1. INTRODUCTION The network simulation is imitation of some network system, and it represents key characteristics or behaviours of a selected system. This can be used in many contexts, for example, performance optimization of a network system, testing devices in the system, etc. Generally, communication system requires large cost to its deployment and it is hard to fix after initial deployment; therefore the network simulation models are introduced. The network simulation models do the simulation using the system s key characteristics and behaviours. The system designer can test the designed network and the system and fix the network before its deployments. Therefore it can reduce system deployment cost. The satellite network
system simulation models are also introduced to reduce system deployment cost, as same as deployment in other communication system. However, some previous satellite network communication simulation models are implemented in the simulator by using special hardware. In addition, they only simulate satellite channel model, therefore it requires other network equipments, such as router or switches, to simulate whole networks. Although other models implemented by software, however these models also focus to simulation of the satellite channel model only. So, it also requires other network equipments as same as hardware version network simulator. Therefore, we started to develop the network simulation model using a software version network simulator to simulate whole satellite network system. 2. THE MF-TDMA SATELLITE NETWORK SIMULATION MODEL 2.1 The Overview of MF-TDMA Satellite Network Simulation Model Figure 1 shows our MF-TDMA satellite network simulation model. The model is implemented using OPNET that is a network simulator. Basically, this model use libraries in OPNET to represent practical protocols such as TCP, UDP, IP, Ethernet, and application protocols, etc. This model also includes new process and node models, whose are our new models, to simulate MF-TDMA system. Moreover, to represent channel and interference model in satellite network system, pipeline model and several attributes are added in the simulation model. The process model is implemented as finite state machine using C language and libraries in OPNET as shown in Figure 2. Also, we implemented pipeline model for representing channel model in the network simulator. Figure 1. MF-TDMA satellite network simulation model
Figure 2. State machine of mobile station in MF-TDMA satellite network simulation model 2.2 Access Control Mechanism The simulation model in our paper is designed to simulate the satellite network system model using Multi Frequency Time Division Multiple Access(MF-TDMA) mechanism[1-2]. We assume that the satellite network system uses MF-TDMA, because it is a candidate for the access control mechanism of the next satellite network system in Korea. In the MF-TDMA, the time and frequency are divided into superframes, which are timefrequency blocks as shown in Figure 3. A superframe contains several frames and they contain a number of slots. A frame can be a clear-sky frame or a rain-fade frame. The clearsky frame is for users in good weather condition region, where as the rain-fade frame is for users in bad weather region. The rain-fade frame has larger slot time(t r f > T c f) and more bandwidth(w r sf > W c sf) than clear-sky frame. A frame contains common signaling channel(csc), acquisition(acq), synchronization(sync), and traffic(trf) slots. In the common signaling channel, users can transmit control message which includes channel request. The synchronization and acquisition slots used to synchronize with control station and find the start of frames. The users transmit their data in allocated traffic slots. In this simulation model, we assumed that downlink channel is time division duplexing(tdd). In the TDD, land earth station, which relays data, control traffic, resource allocation information, and control messages.
Frequency T sf W sf Superframe W c sf W r sf T c f Frame Frame Frame Frame T r f Time Clear-sky frame Rain-fade frame Time C S C A C Q SY NC T R F T R F Figure 3. Superframe and frame structure of MF-TDMA T R F T R F Time 2.3 Channel Models in our Simulation Model Satellite communication system uses satellite on the sky therefore it experiences different environment compared with practical wireless network such as Wireless LAN or cellular networks. So, several researchers study about the environment for satellite communication system and they introduce several channel models[3]. Basically, the signal power in the network decades as the distance between transmitter and receiver is increased. This phenomenon is referred as path loss. In the satellite network, path loss is modeled as Friis path loss model. In this model, path loss L is calculated as (1). 2 L (1) 4 d The atmosphere and ionosphere layer of the earth makes more signal attenuation. This phenomenon is referred as the atmospheric absorption and the ionospheric effect. The atmospheric absorption is very small around 10GHz frequency band therefore it can be treated as constant attenuation. However, ionospheric effect makes fast fading to the signal. Rain also can make signal attenuation. This phenomenon is called as the rain fade. The rain fade is slow fading with fading depth, which related with rainfall rate of region at the transmitter/receiver on the earth, and elevation angle. [4] gives the average fading depth with respect to the rainfall rate of user s region, while [5] provides the dynamics of rain fading.
Neighbor satellites communication can give interference to a satellite communication system, and it decreases the signal-to-interference-noise ratio(sinr) of receiver of that communication system. 3. TESTING THE SIMULATION MODEL 3.1 Simulation Environments To ensure the correctness of the simulation model, the testing and comparison with expected or known results are required. We test our simulation model using expected result of MF- TDMA system and file transmission using FTP. In the following simulation, we set up three computers as shown in Figure 4, whose are the client, network simulation model and server, from left to right, respectively. Also, we use the parameters in the simulation model as shown in Figure 5. The antenna gains are configured to represent Korean satellite system s antennas. Downlink and Uplink data rate is selected for the network system with 6 users. Channel model related parameters are set to represent the position of users in the Korea and geo-stationary earth orbit satellite. Figure 4. Satellite network simulation model(center) connected with client(left) and server(right)
Figure 5. Simulation Environments 3.2 Test using Expected Result of Access Control Mechanism Typical MF-TDMA system periodically receives requests for resource, allocates the resource, and notifies the allocation result to user nodes. If the system does not have logon session maintenance function, then the user nodes using MF-TDMA must request the resource when it has some traffic to transmit. In this situation, the nodes will experience long delay which includes resource allocation time and propagation delay for control messages as shown in Figure 6. If the logon session maintenance function is enabled, then the mobile station and earth station send Ping Request and Ping Response message without re-sending Channel Request and Channel Allocation message. Therefore the end-to-end delay decreases when the logon session maintenance function is enabled.
1. Ping request 2. Channel Request 3. Channel Allocation 4. Ping request 5. Ping request Client 8. Ping response Mobile Station 7. Ping response Earth Station 6. Ping response Server Figure 6. Ping message exchange procedure when logon session maintenance function does not worked Figure 7. Pinging results when logon session maintenance functions is disabled Figure 8. Pinging results when logon session maintenance functions is enabled
Table 1. Round-trip time comparison with respected to the session maintenance function Session Maintenance Function Average round-trip time(ms) Enabled 854.33 Disabled 2609.07 Figure 7 shows the pinging results when logon session maintenance function is disabled. It shows that round-trip time is about 3 seconds as described. Figure 8 shows the pinging results when logon session maintenance function is enabled. It shows that logon session maintenance function makes shorter round-trip delay because of elimination of allocation request and response. Table 1 shows the average round-trip time of the network system. It shows that logon session maintenance function makes shorter round-trip time because of elimination of allocation request and response, as expected. Therefore, our network simulation can be said that this model contains a characteristic of MF-TDMA system. 3.3 Throughput with respect to frame length In the MF-TDMA system, frame length determines system throughput. The amount of wasted resources at the end of allocated resource of one user increases as frame length decreases. For example, if a packet is larger than remained resources then it cannot be transmitted in the current frame and this remained resource is not used by allocated user. These wasted resources decrease the system throughput. We check that whether this phenomenon occurs or not in our simulation model. In this examination, six users generate UDP traffic with amount of 2Mbps each, and send it to the server. There are two users with 100 slots, two users with 200 slots, one user with 300 slots, and one user with 500 slots, over 1,000 slots per frame. Maximum uplink data rate per channel is 1Mbps Figure 9 shows the resource use efficiency of the system with respect to the frame length, while Figure 10 shows the uplink throughput of that system. As expected, the throughput increases as the frame length increases because large frame length makes large slot length, and users can send more packets using same resources. Therefore, our network simulation can be said that this model contains and represents characteristics of MF-TDMA system. 1 Resource Use Efficiency 0.98 0.96 0.94 0.92 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Frame Length(seconds) Figure 9. Resource use efficiency of MF-TDMA system with respect to frame length
7.5 x 105 MF-TDMA Uplink Throughput 7 6.5 6 5.5 5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Frame Length(seconds) Figure 10. MF-TDMA throughput of the system with respect to frame length(unit : bytes/s) 3.4 File Transmission Test using FTP Testing by file transmission using FTP can be a method to verify the simulation model. TCP has maximum throughput limited by round-trip time and maximum window size, which one is network metric and the other is parameter of TCP. The maximum throughput of file transmission is limited as (2). (Window Size) (maximum throughput) (2) (Round-trip Time) Figure 11 shows the transmission status window of a Smart FTP client program which downloading a 5Mbytes file. The client and a file server are connected to the simulation model using system-in-the-loop function of OPNET. The frame size of MF-TDMA system is 50 ms which is required to use TCP. The user, which is the FTP client, allocated 800 slots over 1000 slots per frame. Figure 11. Dialog window of FTP client in our testing Table 2. Measured data rate in the file transmission using FTP File size (Megabytes) Average data rate (kilobytes/seconds) Maximum data rate (kilobytes/seconds) 5.00 52.8 76.7 Table 2 shows the average and maximum data rate during file transmission. As shown in Table 2, average downloading speed is about 52.8kilobytes/seconds, and the maximum data rate is 76.7 kilobytes/seconds. The maximum possible throughput calculated by the TCP window size and end-to-end delay is about (65,535 bytes/854.33 ms) = 76.709 kilobytes/second. Average downloading speed is less than maximum throughput because of following reasons. First, uplink channel use MF-TDMA, therefore traffic experiences variable delay with average delay of 854.33ms. Second, FTP client and FTP server cannot serve maximum throughput because of the overheads of network system. Third, the client doesn t assign full time of a channel and it limits the throughput.
In this test, we can confirm that this simulation model represents the end-to-end characteristics of the satellite network model using geo-stationary earth orbit satellite. It also shows that our MF-TDMA simulation model can be used with real devices; therefore it can be used for developing and debugging network devices in the satellite network system. 4. CONCLUSIONS This paper introduces the MF-TDMA based satellite network system simulation model. The model contains process models and parameters which represents the network system using geo-stationary earth orbit satellite and MF-TDMA access control mechanism. To validate our network simulation model, we expects the possible phenomenon of MF-TDMA based system and geo-stationary earth orbit satellite systems such as end-to-end delay with respect to the session maintenance function, throughput with respect to the frame length of MF-TDMA, and maximum throughput of TCP. These results show that our network system model appropriate to use in the simulation of the geo-stationary earth orbit satellite and MF-TDMA access control mechanism based network system. ACKNOWLEDGEMENTS This work has been supported by National GNSS Research Center program of Defense Acquisition Program Administration and Agency for Defense Development. REFERENCES [1] Ki-Dong Lee and Kun-Nyeong Chang, A Real-Time Algorithm for Timeslot Assignment in Multirate Return Channels of Interactive Satellite Multimedia Networks, IEEE Journal on Selected Areas in Communications, vol. 22, no. 3, April 2004. [2] N.G.V. Kumar, et. al, Digital Video Broadcast Return Channel via Satellite(DVB-RCS) Hub for Satellite based e-learning, The International Journal of Multimedia & Its Applications, vol.3, no.1, February 2011. [3] John. S. Seybold, Introduction to RF Propagation, Wiley-Interscience, 2005. [4] A. L. Johnson, The effects of Ionospheric Scintillation on Satellite Communications, Air & Space Power Journal, Oct. 2003. [5] Specific Attenuation Model for Rain for Use in Prediction Methods, ITU-R Recommendation P.838-3, Int. Telecommunication Union, Geneva., 2005. [6] S. K. Shin, K. Lim, K. Choi, and K. Kang, Rain Attenuation and Doppler Shift Compensation for Satellite Communications, ETRI Journal, vol.24, no.1, pp.31-42, February 2002. [7] S.Lin and D.J. Costello, "Error Control Coding," 2nd edition, Prentice Hall, June, 2004. [8] S. Poonam and T. K. Bandopadhyaya, "Rain Rate Statistics and Fade Distribution of Millimeter Waves in Indian Continents," International Journal of Infrared and Millimeter Waves, Vol. 19, No. 3 / March 1998.