Bluetooth Networking
9.5 Collision Avoidance Scheduling in Bluetooth Access Points
The proliferation of Bluetooth-enabled personal-computing, communication, and mechatronic devices could fuel the demand of Bluetooth access points that enable these devices to access the Internet. Typically, a Bluetooth access point is connected to the Internet via wireline facility such as a local area network (LAN). According to the Bluetooth specification [1], each piconet supports a raw data rate of one megabits per second and a maximum of seven active slaves. Thus, a Bluetooth access point, acting as a master and equipped with one radio, only can provide a maximum of one megabits per second access bandwidth and seven simultaneous connections. Use of multiple radios in a Bluetooth access point can be an effective option to overcome such a limitation [22]. An example of a high-capacity Bluetooth access point with four Bluetooth radios which function as masters of four piconets is shown in Figure 9.8. Here “Pi” represents the i-th piconet with its master located at the access point, and “R” is the radius of the radio coverage, usually around 10 m, of the master nodes. However, the throughput of the Bluetooth access point in Figure 9.8 is less than four times that of an isolated piconet, due to collisions of transmissions FIGURE 9.6 Aggregate throughput (on–off ratio = 2:1).
FIGURE 9.7 Average delay (on–off ratio = 2:1).
0 200 400 600 800 1000 1200 1400 1600
Traffic ratio (beta)
Aggregate throughput (kbps)
ASA CBS FSS
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
0 1 2 3 4 5 6 7 8 9
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Traffic ratio (beta)
Average delay (s)
FSS CBS ASA 0775_C009.fm Page 9 Tuesday, September 4, 2007 9:32 AM
9-10 Mechatronic Systems: Devices, Design, Control, Operation and Monitoring
between piconets that hop onto the same frequency channel. The collision problem becomes more severe when the number of collocated piconets and traffic load increase [22,23].
Bluetooth employs FHSS transmissions over 79 channels each 1 MHz wide. The pseudo-random hop sequence of each piconet is determined by the master node, and all slave nodes are synchronized to it. In the United States, FCC Part 15 rules, which apply to license-free devices operating in the ISM band, do not allow the frequency-hop sequences of different piconets to be coordinated. Similar rules apply in other jurisdictions. However, as all the radios in the Bluetooth access point can share a common baseband module, it is possible to align the time slot boundaries of all the piconets by applying a common clock to all the master nodes in the Bluetooth access point. This approach effectively eliminates collisions that spill into adjacent time slots. Furthermore, as the multiple Bluetooth radios share a common antenna, when one of the radios is transmitting, other radios could become saturated if they are receiving. Therefore, not only should the piconets align their time slots, but they also should align their use of time slots for master–slave and slave–master transmissions, so that the entire Bluetooth access point is either transmitting or receiving in any given time slot. As all transmissions in a piconet are scheduled by the master node, and all master nodes in the Bluetooth access point employ a common baseband module, it follows that the baseband module can schedule transmissions in all the corresponding piconets. Furthermore, as the baseband module has knowledge of all the (independently chosen) hop sequences employed by all the piconets, one can design a collision avoidance scheduling (CAS) algorithm for deployment in the baseband module to prevent transmissions over different piconets from interfering with each other.
The CAS algorithm employs a frequency occupation table in the baseband module to keep track of the usage of the 79 channels. This is necessary as transmissions could span 1, 3, or 5 time slots during each hop. When several piconets attempt to hop to a specific channel, one of the piconets (the “winning piconet”) is selected at random as the candidate for accessing the channel, whereas other piconets defer access to the next hop. The frequency occupation table is next consulted to see if the channel is already occupied by an ongoing transmission in another piconet. If so, the winning piconet defers channel access until the next hop. If no ongoing transmission is found in the channel, the winning piconet is allowed to access the channel, and the frequency occupation table is updated to reflect the number of slots in which the winning piconet will occupy the channel.
We have evaluated system performance through simulations, in which we consider SCO and ACL links with DH1, DH3, and DH5 packets, which occupy 1, 3, and 5 slots, respectively. A typical set of results is shown in Figure 9.9, which compares the system throughput of a Bluetooth access point employing CAS with that without CAS. It can be seen that CAS gives throughput improvements that increase with the number of piconets, i.e., the number of collocated Bluetooth radios at the Bluetooth access point.
Whereas the performance improvements are relatively small for DH1 packets, the amount of improve-ments increases as longer packets are used. With 10 Bluetooth radios in the Bluetooth access point and DH5 packets, CAS gives a throughput improvement of better than 1.5 Mb/s or about 20% over that of a Bluetooth access point without CAS.
FIGURE 9.8 A Bluetooth access point with four piconets.
P1 Access point coverage
P2
P4
P3
Master Active slave Inactive slave
High Capacity Bluetooth access point R
0775_C009.fm Page 10 Tuesday, September 4, 2007 9:32 AM
Bluetooth Networking 9-11
9.6 Conclusion
Bluetooth technology is applicable in communication among mechatronic devices and within a com-plex mechatronic system. In this chapter, we presented some highlights of the Bluetooth networking research completed at UBC under the “Enabling Technologies for Ubiquitous Personal Area Network-ing” project, which considered large-scale Bluetooth networking in ad hoc networking and access-point networking scenarios. For Bluetooth ad hoc networking, we have addressed the scatternet formation problem and intra-piconet and inter-piconet scheduling problem. We have presented our proposed TPSF and TPSF+ algorithms. TPSF+ can obtain more recent on-demand scatternet route information than the original TPSF in a dynamic environment with node mobility. Simulation results show that TPSF+ has a higher aggregate throughput and lower end-to-end delay when compared with Bluenet. We have also described the use of mobile agents in the Bluescouts scatternet formation method, which facilitate organic growth of a scatternet with reconfiguration for topology optimization accom-plished via mobile processing. Results show that large-scale scatternets formed by Bluescouts have a slave-to-master ratio that approaches the theoretical maximum value. For Bluetooth scheduling, we have proposed an ASA for Bluetooth scatternets. ASA can dynamically allocate bandwidth to each link based on traffic demands while maintaining max–min fairness for all the nodes within the scatternet.
Besides, ASA can also prevent bridge node conflicts. Simulation results show that ASA can maintain a high aggregate throughput and low delay on bursty on–off UDP traffic when compared with FSS and CBS. For Bluetooth access point networking, we have considered the frequency hop collision problem in high-capacity Bluetooth access points employing multiple Bluetooth radios, and proposed a collision-avoidance scheduling algorithm that effectively addresses this problem to give improved system throughput.
Acknowledgment
This chapter summarizes the research contributions of research associates Yoji Kawamoto of Sony Corporation during his visit to UBC, Dr. Zhifeng Jiang, Dr. Qixiang Pang, and graduate students Chu Zhang, Sergio González-Valenzuela, and Raymond Lee. Their work was partially supported by the Canadian Natural Sciences and Engineering Research Council (NSERC) under grant number STPGP 257684-02.
FIGURE 9.9 Throughput versus number of piconets (ACL link).
1 2 3 4 5 6 7 8 9
Number of piconets (ACL link)
Overall throughput (Mbps)
DH5 original DH5 CAS DH3 original DH3 CAS DH1 original DH1 CAS
2 3 4 5 6 7 8 9 10
0775_C009.fm Page 11 Tuesday, September 4, 2007 9:32 AM
9-12 Mechatronic Systems: Devices, Design, Control, Operation and Monitoring
References
1. Specification of the Bluetooth System, version 2.0, November 2004, available at http://www.blue-tooth.org.
2. The IEEE 802.15 WPAN Task Group 1a, http://www.ieee802.org/15/pub/TG1a.html.
3. Jain, M., Lampe, L., and Schober, R., Sequence detection for Bluetooth systems, in Proceedings of IEEE Globecom’04, Dallas, TX, November 2004.
4. Lampe, L., Schober, R., and Jain, M., Noncoherent sequence detection receiver for Bluetooth systems, IEEE Journal on Selected Areas in Communications, Vol. 23, No. 9, 1718–1727, September 2005.
5. Lampe, L., Jain, M., and Schober, R., Improved decoding for Bluetooth systems, IEEE Transactions on Communications, Vol. 53, No. 1, 1–4, January 2005.
6. Wu, J.C.H., Aken’ Ova, V., Wilton, S.J.E., and Saleh, R., SoC implementation issues for synthesizable embedded programmable logic cores, in Proceedings of IEEE Custom Integrated Circuits Conference, Santa Clara, CA, September 2003, pp. 45–48.
7. Wilton, S.J.E., Kafafi, N., Wu, J.C.H., Bozman, K., Aken’Ova, V., and Saleh, R., Design consider-ations for soft embedded programmable logic cores, Solid State Circuits Journal, Vol. 40, No. 2, 485–497, February 2005.
8. Law, C., Mehta, A.K., and Siu, K.-Y., Performance of a new Bluetooth scatternet formation protocol, in Proceedings of ACM Symposium on Mobile Ad Hoc Networking and Computing, Long Beach, CA, October 2001, pp. 183–192.
9. Tan, G., Miu, A., Guttag, J., and Balakrishnan, H., An efficient scatternet formation algorithm for dynamic environments, in Proceedings of IASTED Communications and Computer Networks (CCN) Conference, Cambridge, MA, November 2002.
10. Zaruba, G.V., Basagni, S., and Chlamtac, I., Bluetrees—scatternet formation to enable Bluetooth-based ad hoc networks, in Proceedings of the IEEE International Conference on Communications (ICC’01), Helsinki, Finland, June 2001, pp. 273–277.
11. Salonidis, T., Distributed topology construction of Bluetooth personal area networks, in Proceedings IEEE INFOCOM, Anchorage, AK, April 2001, pp. 1577–1586.
12. Basagni, S. and Petrioli, C., A scatternet formation protocol for ad hoc networks of Bluetooth devices in Proceedings of IEEE Vehicular Technology Conference (VTC-Spring), Birmingham, AL, May 2002, pp. 424–428.
13. Wang, Z., Thomas, R., and Haas, Z., Bluenet—a new scatternet formation scheme, in Proceedings of the 35th Annual Hawaii International Conference on System Sciences (HICSS’02), Honolulu, Hawaii, January 2002.
14. Kawamoto, Y., Wong, V., and Leung, V., A two-phase scatternet formation protocol for Bluetooth wireless personal area networks, in Proceedings of IEEE Wireless Communications and Networking Conference (WCNC), New Orleans, LA, March 2003.
15. Zhang, C., Wong, V.W.S., and Leung, V.C.M., TPSF+: a new two-phase scatternet formation algorithm for Bluetooth ad hoc networks, in Proceedings of IEEE Globecom’04, Dallas, TX, November–December 2004, pp. 3599–3603.
16. González-Valenzuela, S., Vuong, S.T., and Leung, V.C.M., BlueScouts—a scatternet formation protocol based on mobile agents, in Proceedings of IEEE ASWN’04, Boston, MA, August 2004, pp. 109–118.
17. González-Valenzuela, S., Vuong, S.T., and Leung, V.C.M., Programmable agents for efficient topol-ogy formation of Bluetooth scatternets, International Journal of Wireless and Mobile Computing (in press).
18. Sapaty, P., Mobile Processing in Distributed and Open Environments, John Wiley & Sons, New York, 2000.
19. Lee, R. and Wong, V., An adaptive scheduling algorithm for Bluetooth ad hoc networks, in Pro-ceedings of IEEE International Conference on Communications (ICC’05), Seoul, Korea, May 2005, pp. 3532–3537.
0775_C009.fm Page 12 Tuesday, September 4, 2007 9:32 AM
Bluetooth Networking 9-13
20. Baatz, S., Frank, M., Kuhl, C., Martini, P., and Scholz, C., Adaptive scatternet support for Bluetooth using sniff mode, in Proceedings of IEEE Conference on Local Computer Networks, Tampa, FL, November 2001, pp. 112–120.
21. Zhang, W. and Cao, G., A flexible scatternet-wide scheduling algorithm for Bluetooth networks, in Proceedings of IEEE IPCCC, Phoenix, AZ, April 2002, pp. 291–298.
22. Lim, Y., Kim, J., Min, S.L., and Ma, J.S., Performance evaluation of the Bluetooth-based public Internet access point, in Proceedings of International Conference on Information Networking 2001, Beppu, Japan, February 2001, pp. 643–648.
23. El-Hoiydi, A., Interference between Bluetooth networks—upper bound on the packet error rate, IEEE Communications Letters, Vol. 5, No. 6, 245–247, June 2001.
0775_C009.fm Page 13 Tuesday, September 4, 2007 9:32 AM
0775_C009.fm Page 14 Tuesday, September 4, 2007 9:32 AM
10-1