1. BLUETOOTH THROUGHPUT IMPROVEMENT USING A SLAVE TO SLAVE PICONET FORMATION By Christophe Lafon and Tariq S Durrani Institute for Communications & Signal Processing Dept. of Electronic & Electrical Engineering University of Strathclyde Glasgow - UK

2. OVERVIEW  Project Aim & Current Work Objective.  Background  Slave to Slave Piconet Formation Overview.  Switching Piconet:  Clock Synchronization.  Frequency Hopping Sequence.  Simulation results.  Conclusion and Further Work.

3. MOTIVATION OF WORK  AIM: New approach to inter-Piconet communication  New policy: SSPF (Slave to Slave Piconet Formation).  OBJECTIVES: To achieve faster jumps between different Piconets.

4. BLUETOOTH BACKGROUND  Operates in the 2.4 GHz unlicensed ISM band.  79 hop frequencies: f = 2402+k MHz, k= 0,..78.  Bandwidth 1Mb/sec with Fast hopping 1600 hops/s  Access Code: AM_ADDR: 3 bits to units to distinguish between Slave unit participating in the Piconet (7 max).  Channel divided into time slot 625  s length  The Package is transmitted in 1, 3 or 5 Slots

5. TIME DIVISION DUPLEX (TDD) TRAFFIC BETWEEN TWO SLAVES S2 S1 S3 MASTER S4 Master Slave 1 Slave 2 Slave 3 Slave 4 time Slots Traffic between Master and Slaves

6. ALTERNATIVE APPROACH: BOTH SLAVES ENTER NEW PICONET  Slave will indicate to Master of an important data transfer to another Slave.  Payload header:  1 for high traffic  0 for low traffic  Both Slaves will then enter in new mode: Slave to Slave Piconet Formation (SSPF).

7. SWITCHING PICONETS  GOAL: To eliminate requirement of guard time within traffic between two Piconets.  Clock Synchronization (Slot misalignment).  Hopping Sequence (Channel Synchronisation).  Meeting time (Synchronize both piconet).

8. CLOCK SYNCHRONISATION  Each Device has a Native Clock (CLKN).  A Piconet is characterized by Master  Frequency Hopping Scheme  Access code (AM_ADDR)  Timing synchronization (CLK)  Master determines the bit rate allocated to each slave  Slaves do not synchronize to the master  Calculate offsets to master’s Bluetooth Clock CLK.  Monitor timing drift

9. SSPF CLOCK SYNCHRONISATION  The new Piconet Clock will not be synchronized with Master(2) Native Clock, but with Master(2) Estimate Clock of Master(1).  Both Piconets will be synchronized according to the initial Master(1).  Master(2) will synchronize to Master(1) with a rendezvous time.

10. MEETING TIME TO AVOID CLOCK DRIFT  A Meeting time is required to readjust the estimated clock CLKE of the New Master(2) to Master(1).  The time delay is about 0.25sec (every 400 slots).

11. HOPPING SEQUENCE  The hopping sequence is transferred from the Master to the Slave during connection Set-up.  The same generated sequence is presented to all devices in Piconet.  New Piconets = new FHS (Frequency Hopping Sequence) created by its Master.

12. SSPF HOPPING SEQUENCE  New Master (slave creating the new Piconet) will decrease each hop frequency by 10 times its address (AM_ADDR) to control the hopping channel (known by all slaves)  Example:  Master(1) FHS: 32, 41, 30, 26, 36, 39  Leading to new FHS:22, 31, 20, 16, 26, 29 Generated by new Master(2) [Ex-Slave (1)]

13. SSPF SCHEDULE MASTER S6 Slave 2 Slave 3 Slave 4 Slave 1 Slave 5 Hopping Sequence Slave 4 Time New Master S1 New Hopping sequence 22 31 20 16 26 21 50 MASTER Slave 1 Slave 2 Slave 3 32 41 30 26 36 31 50 Slave 5 Master 1 Packet Transmission Master S6 Packet Transmission Packet Reception

14. ADVANTAGE OF SSPF  Slaves have two bandwidths to transfer heavy traffic data.  They could easily switch from one Piconet to another at every slot due to synchronisation between the two Piconets.  In case of transmission failure, the Master to Master communication will allow any Master to forward data and continue data transmission.

15. TOTAL THROUGHPUT OF GENERATING PACKETS Total Throughput using SSPF Total Throughput using SPF 1 1.5 2 2.5 3 3.5 4 300 350 400 450 500 550 600 650 700 (641) Total Throughput using SSPF Total Throughput using SPF 1 1.5 2 2.5 3 3.5 4 300 350 400 450 500 550 600 650 700 Simulation Time (sec) Throughput [ Kbit /s] (641) 330 508.6 592.35 621.56 635.2 641 300 340 380 420 460 500 540 580 620 660 123456 Number of Slaves Sharing bothPiconetswith throughput Percentage Improvement Total Throughput [ Kbits /s] 94%92.5%88%79%54%0% 330 508.6 592.35 621.56 635.2 641 300 340 380 420 460 500 540 580 620 660 123456 Number of Slaves Sharing bothPiconetswith throughput Percentage Improvement Total Throughput [ Kbits /s] 94%92.5%88%79% 54% 0%

16. CONCLUSIONS & FURTHERS WORK  SSPF is designed to facilitate inter-Piconet scheduling.  Our simulation shows that inter-Piconet communication could improve the data traffic transfer by > 90%.  Increases fluidity (packet delays) and on less transfer failure.  The Work presented is the 1 st approach of scatternet algorithm and in future will be developed for more than 8 devices.

17. Thank you for your attention Christophe@spd.eee.strath.ac.uk