Department of Information Engineering University of Padova, Italy On the Impact of Fading and Inter-piconet Interference on Bluetooth Performance A note on the use of these ppt slides: We’re making these slides freely available to all, hoping they might be of use for researchers and/or students. They’re in PowerPoint form so you can add, modify, and delete slides (including this one) and slide content to suit your needs. In return for use, we only ask the following: If you use these slides (e.g., in a class, presentations, talks and so on) in substantially unaltered form, that you mention their source. If you post any slides in substantially unaltered form on a www site, that you note that they are adapted from (or perhaps identical to) our slides, and put a link to the authors webpage: Thanks and enjoy!

Department of Information Engineering University of Padova, Italy WPMC02 WPMC02 Honolulu, Hawaii, October 27-30, 2002 On the Impact of Fading and Inter-piconet Interference on Bluetooth Performance Andrea Zanella, Andrea M. Tonello, Silvano Pupolin

Department of Information Engineering University of Padova, Italy WPMC02 WPMC02 Honolulu, Hawaii, October 27-30, 2002 A note on the use of these ppt slides: We’re making these slides freely available to all, hoping they might be of use for researchers and/or students. They’re in PowerPoint form so you can add, modify, and delete slides (including this one) and slide content to suit your needs. In return for use, we only ask the following: If you use these slides (e.g., in a class, presentations, talks and so on) in substantially unaltered form, that you mention their source. If you post any slides in substantially unaltered form on a www site, that you note that they are adapted from (or perhaps identical to) our slides, and put a link to the authors webpage: Thanks and enjoy!

October 27-30, 2002WPMC02 Honolulu Hawaii4 Outline of the contents Bluetooth basic Motivations & Purposes System model Results Conclusions

October 27-30, 2002WPMC02 Honolulu Hawaii5 What the standard says… Bluetooth specifications

October 27-30, 2002WPMC02 Honolulu Hawaii6 Main Characteristics Radio Band:  ISM (2.45 GHz)‏ Available (almost) worldwide Royalties-free: cost saving Interference Immunity  Interference in ISM band has different strength and nature: baby monitors, microwave ovens, lighting devices, WLAN,…  Frequency Hopping Spread Spectrum (FHSS) modulation Since most radio systems are band-limited, it is easy to finding a part of the 80-MHz-wide ISM band with low interference FH_CDMA provides a good multiple access scheme Modulation Scheme  Signal Bandwidth of 1MHz (Frequency Hopping)‏  GFSK (K=0.3): robustness, demodulation is easy, low-cost radio units

October 27-30, 2002WPMC02 Honolulu Hawaii7 Bluetooth piconet Two up to eight Bluetooth units sharing the same channel form a piconet In each piconet, a unit acts as master, the others act as slaves Channel access is based on a centralized polling scheme active slave master parked slave standby slave1 slave2 slave3 master

October 27-30, 2002WPMC02 Honolulu Hawaii8 FH & TDD Each piconet is associated to frequency hopping (FH) channel  The pseudo-random FH sequence is imposed by the master  Time is divided into consecutive time-slots of 625  s  Each slot corresponds to a different hop frequency Full-duplex is supported by Time-division-duplex (TDD)  Master-to-slave (downlink) transmissions start on odd slots  Slave-to-Master (uplink) transmissions start on even slots 625  s t t master slave f(2k)‏f(2k+1)‏f(2k+2)‏

October 27-30, 2002WPMC02 Honolulu Hawaii9 ACHEC access codepacket headerpayload CRC General Packet Format Access Code (AC)‏  All packet exchanged in a piconet have the same AC  Packets that don’t satisfy AC test are immediately discarded Packet Header  Contains, among other information, slave active member receiver address, ARQ flags, payload format, header checksum field (HEC)‏  If the HEC test fails, the packet is immediately discarded Payload  If the CRC test fails, the packet is negative acknowledged PAYL

October 27-30, 2002WPMC02 Honolulu Hawaii10 Multi-slot packets f(k)‏ 625  s f(k+1)‏f(k+2)‏f(k+3)‏f(k+4)‏ f(k+3)‏f(k+4)‏f(k)‏ f(k+5)‏ Packets can extend over one, three or five consecutive slots Carrier frequency remains unchanged Multi-slot packets reduce bandwidth losses due to header & guard time (  220  )‏ Max PAYL size 216 bit 1464 bit 2712 bit

October 27-30, 2002WPMC02 Honolulu Hawaii11 ACL data packet formats ACL: Asynchronous data packet formats  1, 3, or 5 slot long  Unprotected or protected by 2/3 FEC Protected packet formats (DM):  medium data capacity  higher protection against errors Unprotected packet formats (DH)  higher data capacity  more subject to errors Erroneous packets are automatically retransmitted (Stop&Wait ARQ)

October 27-30, 2002WPMC02 Honolulu Hawaii12 Aims of the work Motivations & Purposes

October 27-30, 2002WPMC02 Honolulu Hawaii13 Motivations Scenario  Massive Bluetooth presence: many independent piconets overlap in the same area Questions  How does inter-piconet interference affect single user performance?  What’s the impact on multi-slot packets?  Is there an easy way to derive performance bounds?

October 27-30, 2002WPMC02 Honolulu Hawaii14 Aims of the work Previous works  Either based on simulations or making restrictive assumptions as Fixed length packets Destructive interference Absence of fading Our contributions  Simple method to evaluate impact of fading and inter-piconet interference on Bluetooth performance taking into account Fading & capture effects Packet format (FEC and packet length)‏ Arbitrary number of potential interferers

October 27-30, 2002WPMC02 Honolulu Hawaii15 Hypothesis and assumptions System Model

October 27-30, 2002WPMC02 Honolulu Hawaii16 Interferers model We focus on a Target Receiver (TR)‏  TR is r 0 meters apart from its transmitter  TR can receive any pck type N p “potential” interferers  Uniformly distributed around TR, s.t. f r (r)=2r/D 2  Use only 1-slot long pcks  Nodes are slot-synchronous N e “effective” interferers per slot r0r0 D Target receiver Interf. Piconet 2 Interf. Piconet 3 Interf. Piconet 1 Time Frequency carriers

October 27-30, 2002WPMC02 Honolulu Hawaii17 Bit Error Probability P rx : received power P I : interference power N 0 : noise power R I : SIR for BER=1e-3 R 0 : SNR for BER=1e-3 * Zürbes et al. “Radio Network Performance of Bluetooth,” ICC 2000

October 27-30, 2002WPMC02 Honolulu Hawaii18 Propagation model Slow Flat Fading (envelope=  )‏  flat on 1 MHz channel  constant along the entire packet  Rice or Rayleigh distributions WSSUS  Signals from different transmitters incur independent fading  Because of FH, successive pcks experiment independent fading For Rayleigh fading, we get Power-addictive interference  N e interferers powers add up to P I

October 27-30, 2002WPMC02 Honolulu Hawaii19 Conditioned PER ACHEC PAYLOAD 72 bits 54 bits h=220  2745 bits CRC Correlator Threshold (CT) ‏ 2-time bit rep. ( 1/3 FEC) ‏ DHn: Unprotected DMn: (15,10) Hamming FEC  0 : BER λ 0 : normal. useful power Λ: normal. interfer. power

October 27-30, 2002WPMC02 Honolulu Hawaii20 Average PER AC HEAD PAYL 220 µs T slot =625 µs · = · = · Target Packet Effective Interferers

October 27-30, 2002WPMC02 Honolulu Hawaii21 Analysis and Simulations Results

October 27-30, 2002WPMC02 Honolulu Hawaii22 Performance Metrics We focus on forward-channel performance:  Packet type Dxk x  {M,H}, k  {1,3,5}  Disregard errors in the backward channel Performance metrics  Average Packet Error Probability: PEP xk Probability of Dxk pck dropping due to unrecoverable errors  Average Forward Throughput:  xk Average number of useful user data bit transmitted in the forward direction per unit of time  Throughput Crossing Point:  N x Number of potential interferers for which x5  x3

October 27-30, 2002WPMC02 Honolulu Hawaii23 Results (1)‏ Parameters  Rayleigh fading (K T =-  dB)‏  r 0 =8 m, D = 10 m  Different pck formats Results  PEP curves for DMk and DHk get close each other as the number Np of potential interferers increases FEC does not give benefit  As expected, Dx3 formats outperform Dx5 in terms of throughput for Np  10

October 27-30, 2002WPMC02 Honolulu Hawaii24 Results (2)‏ Parameters  Interferers: Rayleigh fading  Target Receiver: Rayleigh: K T =-  dB Ricean: K T =6 dB Results  The crossing point Nx is greater than 10 when transmitter an receiver are within the nominal coverage range (10 m)  The presence of LOS has a marginal impact on Nx, but the throughput at the crossing point is higher Shorter formats are better

October 27-30, 2002WPMC02 Honolulu Hawaii25 Analysis Accuracy Analysis vs Simulations  Simulator does consider time offsets, fading, … Results   =( theor - sim )/ sim  Bound is tight for Np 7  Bound is loose for Np>10, r 0  5

October 27-30, 2002WPMC02 Honolulu Hawaii26 Conclusions In case of inter-piconet interference  DMn formats achieve very poor performance  Dx5 and Dx3 formats show a performance tradeoff for increasing number of interferers  The performance crossing point Nx depends on the distance r 0 between transmitter and receiver and marginally on the presence of LOS  In case of LOS, short formats may be more suitable than long ones Model Accuracy  Analytic bound becomes loose for high number of potential interferers and intermediate values of r 0

October 27-30, 2002WPMC02 Honolulu Hawaii27 Thanks! Thanks for you attention!