Scenario 2: Heavy Offered Load

Table 4.2: Properties of the data flows used in scenario 2.

There are three data flows in scenario 2: two FTP flows and one exponential distributed data flow (exponential traffic). The FTP flow simulates bulk data transfer and will occupy as much as bandwidth as possible (ABR). The specifications of scenario 2 are listed in Table 4.2. In this scenario, the offered load was close to 100%

of the total capacity and the link utilization was almost full all the time. That is, the throughput degradation is more susceptible by increasing the number of interference-affected channels. We also set r=0.5 for QSD-PR and compare the link utilization and throughput with different Ratio^threshold in scenario 2.

Fig. 4.7 shows that the link utilization when the percentage of interference-affected channels increase from 0% to 100%. When the percentage of interference-affected channels increases from 0% to about 40%, the QSD-PR with CSD-SAR could still maintain high link utilization and had high throughput. This is because the QSD-PR with CSD-SAR uses multi-slot packets to mask interference-affected channels. However, when the percentage of interference-affected

Property

Slave no. Traffic Type Data Rate

Transport

Layer Packet Size Burst Time

channels increases over 40% of the total channels, the link utilization decreases gradually because of delayed transmission. In Fig. 4.8, it shows that by applying our QSD-PR with CSD-SAR scheme, we can achieve higher throughput in either error-free or error-prone environments. Note that the threshold value Ratio^threshold in CSD-SAR can be optimized and further enhance the throughput.

0 10 20 30 40 50 60 70 80 90 100

Link utilization (% of total slots)

Interference-affected channels (%) QSD-PR+CSD-SAR (Ratio^threshold = 2)

QSD-PR+CSD-SAR (Ratio^threshold = 1) QSD-PR+CSD-SAR (Ratio^threshold = 0.5) RR+R-SAR (DH)

RR+R-SAR (DM)

Fig. 4.7: Link utilization vs. interference-affected channels (%) in scenario 2.

In scenario 2, the performance of our proposed scheme is significantly better than the RR with R-SAR scheme because we use the receiving frequency table to avoid bad channels and we fragment the transport layer packets as large as possible to mask bad frequencies by using multi-slot packets. In addition, using the DM and DH packets based on channel conditions can efficiently reduce the packet error rate and guarantee high throughput in error-prone environments. Simulation results show that the QSD-PR with CSD-SAR scheme can adapt to error-prone environments under high load.

0 10 20 30 40 50 60 70 80 90 100 QSD-PR+CSD-SAR (Ratio^threshold = 0.5)

QSD-PR+CSD-SAR (Ratio^threshold = 1) QSD-PR+CSD-SAR (Ratio^threshold = 2) RR+R-SAR (DH)

RR+R-SAR (DM)

Fig. 4.8: Throughput vs. interference-affected channels (%) in scenario 2.

According to the results from Fig. 4.8, we set Ratio^threshold =0.5 and gave the mean state residency time X and _G X_B with different values and observed the throughput improvements using the proposed scheme. In Fig. 4.9, it shows that our QSD-PR with CSD-SAR scheme can offer throughput improvements as high as 195%

compared to the RR with R-SAR (DH) scheme (when the percentage of interference-affected channels increases from 0% to 70%). In addition, the QSD-PR with CSD-SAR allows the Bluetooth system to remain usable even when all the channels are interference-affected. Note that when channels are more error prone, the more improvement can be obtained by using our proposed scheme.

Finally, we investigate the effect of number of slaves on throughput improvements by increasing the number of slaves from 3 slaves up to 7. The traffic type of slaves 3 through 7 is the exponential traffic that specified in Table 4.2. A piconet has a limit on the maximum number of active slaves. In Fig. 4.10, by

increasing the number of slaves in the piconet, we can see that the more throughput improvements can also be obtained by using our proposed scheme. This is because each slave in the piconet has a different data input rate and the RR scheduling scheme will waste more baseband slots by polling sources with low data input rates. It results in lower link utilization and thus lower throughput using the RR.

0 10 20 30 40 50 60 70 80 90

Fig. 4.10: Throughput improvements with various numbers of slaves in a piconet.

Chapter 5

Conclusions and Future Work

5.1 Concluding Remarks

The market is rapidly moving toward resolving the coexistence concerns surrounding the IEEE 802.11b and Bluetooth [31]. Our proposed approach and other approaches have addressed the issue before it ever affects the end-user. As a result, market forecast for Bluetooth and IEEE 802.11b will remain strong, and the need for effective, multi-standard, coexistence solutions will only increase as wireless devices proliferate and simultaneous operation usage models become pervasive [31].

Simulation results have shown that out packet selection and scheduling scheme based on the channel state and queue state can have higher link utilization and higher throughput compared to the Round Robin packet scheduling scheme in an interference environment. In addition, the scheduling policy that delays transmission to avoid bad frequencies occupied by other devices will alleviate the impact of interference on the other systems significantly [8]. Note that our scheme can also be adapted and used in other centrally controlled TDD wireless systems, such as IEEE 802.15.1.

5.2 Future Work

We will look for additional scenarios for a variety of traffic sources and take the SCO link into consideration and study the performance of real-time applications (e.g. voice) using our proposed packet selection and scheduling scheme. In addition, the parameters of QSD-PR and CSD-SAR, such as r and Ratio^threshold, will be further optimized by mathematical analysis or simulation.

Bibliography

[1] The Bluetooth Web Site, http://www.bluetooth.com.

[2] Bluetooth Special Interest Group, Specification of the Bluetooth System, Version 1.1, Feb. 2001.

[3] S. Kapp, “802.11: leaving the wire behind,” IEEE Internet Computing, vol. 6, issue 1, pp. 82-85, Jan.-Feb. 2002.

[4] K. J. Negus, A. P. Stephens, and J. Lansford, “HomeRF: Wireless networking for the connected home,” IEEE Personal Communications, vol. 7, issue 1, pp. 20-27, Feb. 2000.

[5] ETSI, “Broadband Radio Access Networks (BRAN): HIPERLAN type 2 technical specification,” April 2000.

[6] J. Lansford, A. Stephens, and R. Nevo, “Wi-Fi (802.11b) and Bluetooth:

enabling coexistence,” IEEE Network, vol. 15, issue 5, pp. 20-27, Sept.-Oct.

2001.

[7] Javier del Prado and Sunghyun Choi, “Experimental study on coexistence of 802.11b with alien devices,” in Proc. of the IEEE 54^th Vehicular Technology Conference, Oct. 2001, vol. 2, pp. 977-981.

[8] N. Golmie and N. Chevrollier, “Techniques to improve Bluetooth performance in interference environments,” in Proc. of IEEE Military Communications Conference, Oct. 2001, vol. 1, pp. 581-585.

[9] N. Golmie, N.Chevrollier and I. EIBakkouri, “Interference aware Bluetooth packet scheduling,” in Proc. of IEEE Global Telecommunications Conference, Nov. 2001, vol. 5, pp. 2857-2863.

[10] S. Deb, M. Kapoor, and A. Sarkar, “Error avoidance in wireless networks using link state history,” in Proc. of IEEE INFOCOM, April 2001, vol. 2, pp. 785-795.

[11] Y.-H. You, et al., “Adaptive frequency hopping scheme for interference-limited WPAN applications,” Electronic Letters, vol. 37, issue 15, pp. 976-978, July 2001.

[12] J. C. Haartsen, “The Bluetooth radio system,” IEEE Personal Communications, vol. 7, pp. 28-36, Feb. 2000.

[13] Bluetooth Special Interest Group, “Bluetooth protocol architecture,” White paper, Aug. 1999.

[14] Bluetooth Special Interest Group, “Bluetooth Networking Encapsulation Protocol (BNEP) Specification,” Specification of the Bluetooth System, Version 0.95, May 2001.

[15] Bluetooth Special Interest Group, “Bluetooth Personal Area Networking Profiles,” Specification of the Bluetooth System, Version 0.95, May 6, 2001.

[16] S. Zurbes, “Considerations on link and system throughput of Bluetooth networks,” in Proc. of the 11^th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, Sept. 2000, vol. 2, pp. 1315-1319.

[17] J. Kim, Y. Lim et al., “An adaptive segmentation scheme for the Bluetooth-based wireless channel,” in Proc. of the 10th International Conference on Computer Communications and Networks, Oct. 2001, pp. 440-445.

[18] M. C. Valenti, M. Robert, and J. H. Reed, “On the throughput of Bluetooth data transmission,” in Proc. of IEEE Wireless Communications and Networking Conference, Mar. 2002, vol. 1, pp. 119-123.

[19] T.Y. Chui, F. Thaler and W.G. Scanlon, “Bit error rate related load constraints for Bluetooth baseband packets,” Electronic Letters, vol. 38, issue 3, pp. 137-138, Jan. 2002.

[20] J. Zander and J. Malmgren, “Adaptive frequency hopping in HF communications,” IEEE Proceedings – Communications, vol. 142, issue 2, pp.

99-105, April 1995.

[21] J. Bray, and C.F. Sturman, Bluetooth: Connect without cables. Prentice Hall PTR, 2001, pp. 348-350.

[22] A. Das, A. Ghose, A. Razdan, H. Saran, and R. Shoery, “Enhancing performance of asynchronous data traffic over the Bluetooth wireless ad-hoc network,” in Proc of IEEE INFOCOM 2001, April 2001, vol. 1, pp. 591-600.

[23] P. Bhagwat, P., et al., “Enhance throughput over wireless LANs using channel state dependent packet scheduling,” in Proc. of IEEE INFOCOM, Mar. 1996, vol.

3, pp. 1133-1140.

[24] C. Fragouli, V. Sivaraman, M. Srivastava, “Controlled multimedia wireless link sharing via enhanced class based queuing with channel state dependent queuing,” in Proc. of IEEE INFOCOM 1998, Mar. 1998, vol. 2, pp. 572-580.

[25] ns Notes and Documentation, the VINT Projects, UC Berkeley, LBL, USI/ISI and Xerox PARC, http://www.isi.edu/nsnam/ns/.

[26] Bluehoc simulator, a Bluetooth extension to NS, http://www-124.ibm.com/developerworks/opensource/bluehoc/.

[27] H. Wang, and N. Moayeri, “Finite state Markov channel – a useful model for radio communications, “ IEEE Trans. On Vehicular Technology, vol. 44, no. 1, pp. 163-171, Feb. 1995.

[28] E. O. Elliott, “Estimates of error rate for codes on burst-noise channels,” Bell System Technical Journal, vol. 42, pp.1977-1997, Sept. 1963.

[29] E. N. Gilbert, “Capacity of burst-noise channel,” Bell System Technical Journal, vol. 39, pp. 1253-1266, Sept. 1960.

[30] M. Kalia, D. Bansal, R. Shorey, “Data Scheduling and SAR for Bluetooth MAC,” in Proc. of Vehicular Technology Conference, May 2000, vol. 2, pp.

716-720.

[31] Mobilian Corp., “Wi-Fi and Bluetooth: An examination of coexistence approaches,” http://www.mobilian.com/documents/WinHEC_whitepaper.pdf.