LAB PROJECT Nº1 1 INTRODUCTION REDES MÓVEIS E SEM FIOS , MEEC PERFORMANCE OF WIRELESS PERSONAL AREA NETWORKS

Transcription

1 LAB PROJECT Nº1 REDES MÓVEIS E SEM PERFORMANCE OF WIRELESS PERSONAL AREA NETWORKS 1 INTRODUCTION The performance of wireless networks is highly dependent of physical phenomena, such as path loss, shadowing, absorption, scattering, diffraction and fading. The shared nature of the wireless medium makes communication even more difficult, since wireless stations must share or compete for the available bandwidth. Transmission errors caused by collisions and/or Signalto-Noise-plus-Interference-Ratio (SNIR) degradation due to interfering devices, are very common in contention based Wireless Local Area Network (WLAN) or Wireless Personal Area Network (WPAN) technologies, such as those based on IEEE or IEEE This lab project aims to study some of the factors that affect WPAN performance by means of simple tests implemented in the OMNET++ simulator. Throughout this document, the students shall be asked to build simulation script files. Although the main guidelines for scenario construction are provided, additional procedures may be necessary, which require initiative from the students. The students shall also be asked to answer questions regarding the simulated network environment and performance results. The answers shall be included in the report, which is due on the first Sunday after the end of the project. Together with the report, the students must deliver all the scripts and spread sheets that resulted from their work. 1 P á g i n a

2 2 WORK DESCRIPTION The fixed wireless access performance analysis shall be done in several steps, in which relevant independent variables shall be analyzed. 2.1 Impact of Distance in Point-to-Point Communications Within the OMNET++ IDE, add the IEEE project to inet/examples/wireless. Modify the example, so that there are only two hosts: host[0] and host[1] (see Figure 1). Since the objective of this experiment is to measure the impact of distance on the communications performance, we need to use a static mobility model allowing us to set the precise locations of the nodes on the scenario map. The required mobility model is the StationaryMobility, instead of StaticConcentricMobility that comes in the original example (see omnetpp.ini). In order to be able to freely assign node positions from the omnetpp.ini file, the initfromdisplaystring attribute of the mobility modules, should be set to false. Node positions are set using the initialx and initialy attributes of the mobility modules. Figure 1: Modified IEEE example. Using the example that was just built, answer the following questions in your report: Q2.1.1) In this example, the used propagation model is the Log-Distance path loss model, which is here called the BrakpointPathLoss model. You can find it in inet/src/inet/physicallayer/pathloss. Annalise BreakpointPathLoss.ned and BreakpointPathLoss.cc, namely the method BreakpointPathLoss::computePathLoss. Match the parameters with the Log-Distance path loss model parameters. Explain how the received power is calculated. Q2.1.2) Based on your answer to Q2.1.1, calculate the maximum communication range (indicated by the blue circle) with the default parameters. How can you control the latter from omnetpp.ini? Note: you may have a look at the BreakpointPathLoss::computeRange function. You 2 P á g i n a

3 may also need some radio parameters defined in inet/src/inet/physicallayer/ieee802154/ Ieee802154NarrowbandScalarRadio.ned. Q.2.1.3) Measure the impact of distance, packet size and use of acknowledgements on: Application throughput (in kbit/s), Application message latency (s) Average Signal-to-Noise-plus-Interference-Ratio (SNIR) Number of missed Acknowledgements per frame transmission attempt, at the MAC layer. Explain the obtained results. The considered distances between host[1] (sender) and host[0] (receiver) should be 5 equally spaced points between 10m and the maximum transmission range. For each distance, three packet sizes should be considered: 10 and 80 octets. Moreovever, for each distance and packet size combination, results should be taken with and without MAC layer ACK. You will need to manipulate parameters of the application (IIPvXTrafficGenerator). You can find its parameter definitions at src/inet/applications/generic/ IIPvXTrafficGenerator.ned. You will also have to turn MAC layer ACK on and off. You can check the MAC layer parameters at src/inet/linklayer/ieee802154/ Ieee802154NarrowbandMac.ned. Note: you should not change the files under the src folder; all simulation configurations are performed from the omnetpp.ini file. The results should be presented in three charts (throughput, latency and SNIR), each with two curves, one curve for each packet size. Each point in the chart should correspond to to the average over 3 trials, using different random seeds, setting the traffic stream duration of the application to 6 seconds 1. The confidence intervals for the points in the chart should be represented as error bars, or presented in a table. Please read Section of the OMNET++ User Manual in order to get more info on how to repeat experiments with different random seeds. Explain the results. The chart represented in Figure 2 is an example of what the intended graphics should look like (note: the curves were invented for illustration purposes and do not conform to actual results). 1 Not to be confused with the overall simulation time limit. 3 P á g i n a

4 Throughput [kbit/s] REDES MÓVEIS E SEM Distance [m] 20 octets 500 octets 1000 octets Figure 2: Example of chart exemplifying throughput as a function of distance between nodes. 4 P á g i n a

5 2.2 Hidden Terminal and Exposed Terminal Problems Within the OMNET++ IDE, duplicate the modified IEEE example, renaming it IEEE802154_cs. Perform the following experiments: Q2.2.1) Setup a network formed by four stations, A, B, C and D, configuring the topology and data flows in such a way as to simulate the exposed terminal problem. Note that the MAC layer acknowledgements must be switched off. Consider packet sizes of 10, and 80 octets. a) What is the average throughput for each packet size (average over 5 trials, traffic generation time of 20 seconds)? Note: the sending application should be configured to generate a data rate that saturates the network. b) How would you solve the exposed terminal problem? Q2.2.2) Setup a network formed by four stations A, B, C and D, placed in a linear topology, where A sends data to B and C sends data to D, all at maximum possible rate. Consider packet sizes of 10 and 80 octets. MAC layer acknowledgements should now be switched on. a) Measure the average throughput and number of collisions for each packet size (average over 5 trials, traffic generation time of 20 seconds), when all the stations are within communications range of each other. b) Repeat a), but at this time, B is within range of A, C is within range of B and D is within range of C. All the other combinations are out-of-range. Explain the results. 5 P á g i n a

6 2.3 Performance of IEEE under Contention Within the OMNET++ IDE, duplicate the IEEE example, renaming it IEEE802154_star. Modify the example so that a variable number of sender hosts are deployed in circle topology around the receiver, host[0]. All the nodes should be within communications range of each other. The incremental angle for each client is 1 degree. Other modifications may be required, which are not specified in this document. Perform the following experiments: Q2.3.1) Measure the average application throughput, as well as the number of missed MAC acknowledgements per MAC frame transmission, for the following numbers of sender nodes: 1, 10, 25, 50 (average over 5 trials, application traffic generation time of 20 seconds). Consider packet sizes of 10 and 80 octets. Explain the results. Note: the sending application should be configured to generate a data rate that saturates the network. Q2.3.2) Using the results from Q.2.3.1, compare the application and MAC throughputs. Explain the found differences. 6 P á g i n a

7 3 GRADING Grading of the responses shall be performed according to Table 1. Table 1: Grading of LAB1 responses. Question Value Q Q Q Q2.2.1.a 3.0 Q2.2.1.b 2.0 Q a 3.0 Q2.2.2.b 2.0 Q Q REDES MÓVEIS E SEM The organization of the delivered report and scenario files contributes with negative points to the final mark. 7 P á g i n a