1. Slide 1 National Research Council - Pisa - Italy Marco Conti Italian National Research Council (CNR) IIT Institute Measurements of IEEE 802.11 in Ad Hoc Configurations MobileMAN Workshop - Sophie-Antipolis 6-7 March, 2003

2. Slide 2 Outline  Measurements of IEEE 802.11 in Ad Hoc configurations  Problems with current 802.11b cards  Discussion on Simulative studies of IEEE 802.11b

3. Slide 3 IEEE 802.11, hidden stations and TCP The behavior of TCP/IP protocols on top of IEEE 802.11 is an open research issue Some simulative studies seem to point out that may exist Capture phenomena and severe Unfairness problems These studies are highly dependent on the interference model Measurements studies are required

4. Slide 4 Exposed node problem BCDA A = Destination B = Source C = Exposed Node

5. Slide 5 Capture effect BCDA While A is communicating with B 1.D is not aware and makes successive attempts (with increasing back off value) to contact C 2.If the maximum number of attempts is achieved, 802.11 gives up 3.D TCP times out 4.The D to C connection slow down due to the TCP congestion control

6. Slide 6 The Transmission Range (TX_Range) represents the range (with respect to the transmitting station) within which a transmitted packet can be successfully received. The Physical Carrier Sensing Range (PCS_Range) is the range (with respect to the transmitting station) within which the other stations detect a transmission. Interference Range (IF_Range) is the range within which stations in receive mode will be “interfered with” by a transmitter, and thus suffer a loss. IEEE 802.11 behavior

7. Slide 7 The following relationship exists between the ranges: TX_Range <= IF_Range <=PCS_Range IEEE 802.11 behavior: Simulative Studies

8. Slide 8 IEEE 802.11 measurements  Configurations Four laptops WaveLAN IEEE 802.11 Lucent Technologies © R-F Frequency Band: 2.4GHz Modulation Technique: DSSS Data Speed: 2Mbit/s Type of experiments: Indoor Experiments Semi-open space, i.e., outdoor with some obstacles

9. Slide 9 IEEE 802.11 measurements Indoor Experiments. In this case the experiments were performed in a scenario characterized by hidden stations

10. Slide 10 Outdoor Experiments. Two ftp sessions are contemporary active. The arrows represent the direction of the ftp sessions. The payload size of TCP packets was set to 512 bytes d(1,2)= d(3,4) < TX_Range, in addition, by varying the distance d(2,3), the couples of nodes are: i)Type 1: all nodes in the same transmitting range ii)Type 2: extreme case out of the same carrier sensing range (Exp#2) iii)Type 3: Intermediate case #1 d(2,3)=65m iv)Type 4: Intermediate case #2 d(2,3)=15m IEEE 802.11 measurements

11. Slide 11 Type 1 and Type 2 Experiments IEEE 802.11 measurements In Type 1 experiments (all stations within the same transmission range) the two ftp sessions fairly share the bandwidth, and the aggregate throughput is close to the reference throughput for this configuration In Type 2 measurements the two sessions are independent, and they both achieve a throughput very close to the reference throughput

12. Slide 12 Type 3 Experiments IEEE 802.11 measurements Unlike the previous ones, Type 3 and Type 4 experiments exhibited a very strange and unpredictable behavior It can be observed that the use of the RTS/CTS mechanism produces a capture of the channel by the second session When using the basic access mechanism, S1 can start transmitting to S2 without almost any interference from session S3-S4.

13. Slide 13 Type 4 Experiments IEEE 802.11 measurements In Type 4 experiments, whose results are shown in, we observed the capture of the channel by one of the two TCP connections. In this case the RTS/CTS mechanism provided a little help in solving the problem

14. Slide 14 Type 4 configuration, but traffic flows are now generated by CBR sources and the UDP protocol is used instead of TCP IEEE 802.11 measurements Type 4 experiments: UDP traffic

15. Experimantal Perfomance Measurements with Wi-Fi (802.11b)

16. Software :  Operative System: Linux Mandrake 8.2  Software for the traffic generation : DBS  Software to trace MAC PDU : Snuffle Experimental Environment 2 1 3 4 Hardware :  Wireless D-LinkAir DWL-650 card ( IEEE 802.11b )  Laptops (4+1) M  Software for supporting Control traffic  visualization (RTS/CTS/Ack): Wireless Extension Physical Environment :  Open-space areas near CNR in Pisa.

17. 802.11 and 802.11b  802.11b enables transmissions at 5.5 Mbps and 11 Mbps, in addition to 1 Mbps and 2 Mbps, still guaranteeing the interoperability with 802.11 cards  To ensure coexistence and interoperability with 802.11 cards, headers and data fields are transmitted at different data rates to ensure the interoperability between 802.11 and 802.11b cards: At the Physical Layer the 144-bit preamble and the 48-bit header are both transmitted at 1 Mbps. Control frames (RTS, CTS and ACK) and Multicast/Broadcast frames are transmitted at 1 or 2 Mbps. Data frame can be transmitted at any of the NIC data rates.

18. 802.11b Theoretical Throughput

19. Theoretical vs. Real Throughput Comparison between the theoretical maximum throughput and the actual throughput achieved by TCP/UDP applications

20. Transmission Range: TX DATA e TX CONTROL IEEE 802.11b DATA Frame: Control Frame : 2 Mbps 11 Mbps 5.5 Mbps 2 Mbps TX CONTROL TX DATA TX CONTROL TX DATA (11 Mbps) TX DATA (5.5 Mbps) TX DATA (2 Mbps) =

21. Measurement of the Transmission Range: TX DATA and TX CONTROL (2) 2 1 TX DATA  Real difference exists between TX DATA and TX CONTROL

22. Impact of TX DATA and TX CONTROL The Communication Gray Zones Problem the AODV protocol updates routing tables based on the neighboring sensing mechanism (i.e., based on the reception of hello message) it may occur that stable and longer routes are replaced by shorter but unreliable ones

23. 4 Nodes (outdoor experiments) i)interdependencies among the stations extends beyond the transmission range; ii)the physical carrier sensing range often produces an effect that is similar to that achieved with the RTS/CTS mechanism (virtual carrier sensing). iii)The difference in the throughputs achieved by the two sessions when using the UDP protocol (with or without RTS/CTS) can be explained by considering the asymmetric condition that exists on the channel: station S3 is exposed to transmissions of station S1 iv)S1 is exposed to S3 transmissions but the S3’s effect on S1 is less marked Network configuration at 11 Mbps TX (11 Mbps) = 30 m Data length = 512 Bytes Reference throughput = 3000 S1S1 S2S2

24. 4 Nodes (outdoor experiments) S1 and S3 are within the same transmission range and, in addition, it can assumed that all stations are within the same physical carrier sensing range. The system is more balanced from the throughput standpoint. This can be expected as by transmitting with a 2-Mbps data rate the transmission range is significantly larger than with the 11-Mbps transmission rate, and hence the stations have a more uniform view of the channel status. Network configuration at 2 Mbps S1S1 S2S2 95 m TX (2 Mbps) = 90-100 m Data length = 512 Bytes Reference throughput = 1.300 MBytes

25. 4 Nodes (outdoor experiments) Throughputs at 2 Mbps Throughputs at 11 Mbps TX (2 Mbps) = 90-100 m Data length = 512 Bytes Reference throughput = 1.3 MBytes TX (11 Mbps) = 30 m Data length = 512 Bytes Reference throughput = 3 MBytes

26. Ongoing/planned experiments Ongoing experiments, also indicates that the TX Range depends also significantly on the height, i.e., distance from the ground We are also trying to investigate the Carrier Sensing and the Interference Range

27. Ad Hoc Networks Simulation SIMULATION environments :  no one assumed different transmission range s  TX (simulation) >> TX (real) real

28. Ad Hoc Networks Simulation (cont.) Typical scenario: 50 nodes moving in a retangular field (1500 x 300 m) Large TX ranges made links more reliable than in a real case: Route recomputation is a rare event

29. Ad Hoc Networks Simulation (cont.) Random Waypoint model users move on a broken line pattern with a pause time at each vetrex and speed randmly chosen in [0, v max ] Several problems have been recently pointed out in this model

30. Ad Hoc Networks Simulation (cont.) C++ e OTcl (Object Tool Command Language) C language Slow executionQuick execution unstableMore stable tracestatistics NS2 GloMoSim Simulation Tools comparison

31. Ad Hoc Networks Simulation (cont.) C++ e OTcl (Object Tool Command Language) C language Slow executionQuick execution unstableMore stable tracestatistics NS2 GloMoSim Simulation Tools comparison

32. Ad Hoc Networks Simulation (cont.) DSR vs. AODV Speed = 1m/s 3 FTP connections

33. Ad Hoc Networks Simulation (cont.) DSR vs. AODV Speed = 20 m/s 3 FTP connections

34. Ad Hoc Networks Simulation (cont.) DSR vs. AODV Speed = 1 m/s 3 FTP connections

35. Ad Hoc Networks Simulation (cont.) 3 FTP connections + N CBR flows

36. Ad Hoc Networks Simulation (cont.) 3 FTP connections + N CBR flows