Transcript Slide 1
CC2420 Channel and RSSI Evaluation COnnecti vity Lab Dept. of EECS, UC Berkeley Nov/22/2006 Contents Introduction Chipcon CC2420 Radio CC2420 RSSI Evaluation Introduction The Comparison of Different Methods RSS values measured by COTS wireless systems are sufficiently rich in information to permit a mobile device locates itself, reliably and accurately The most important trade off is between Offline Calibration and accuracy The CDF of Different RSS Based Indoor Positiong Methods (Note: Different Experiment Environments) 1 0.9 0.8 0.7 CDF 0.6 0.5 0.4 MP&AP Perspective + Kalman Filter Histogram Bayseian Approach Adaptive Temporal Novel ToF(Steven Lanzisera - UC Berkeley) Hours (Cluster&Cluster Key) Ekahau System Particle Filter (Median error) A System for LEASE (Median error) ECOLOCATION (Max error) 0.3 0.2 0.1 0 0 1 2 3 4 Error (m) 5 6 7 Last Final Conclusion Next Steps Continuing The Project Sensors Evaluation + Fingerprinting Distributed Estimation (Study, Modeling, Simulation, Implementation) Radio Map Independent Algorithm (Study, Modeling Simulation, Implementation) Tx1 Rx3 Rx1 Rx2 Last Final Conclusion The Goal of this Part of the project Become familiar with our sensors and evaluate them as much as possible Understand the behavior of RSSI in indoor environments Measure the effect of different parameters on RSSI Survey the existence of correlation in RSSI Samples and base band received signals Performed Works Study the theory of indoor wireless channels (as much as it was needed) Gathering a lot of data in different situations and environment Analysis of these data Chipcon CC2420 Radio The Chipcon CC2420 Radio Chip Characteristics The CC2420 The CC2420 is an IEEE 802.15.4 compliant RF transceiver designed for low-power and low-voltage wireless applications Frequency: 2.4 GHz Communication Technology: Direct Sequence Spread Spectrum Total BW: 83 MHz Data Rate: 250 kbps Output power: Programmable (Max Power = 0 dBm) Range (Indoor) : 30 m Chipcon CC2420 Radio The Chipcon CC2420 Radio Chip Characteristics The CC2420 IF: 2 MHz Each Byte 2 Sym: 4 Bits 32 Chips Tc = 0.5 us Pulse Shaping: Half Sine Chipcon CC2420 Radio The Chipcon CC2420 Radio Chip Characteristics RSSI and LQI in The CC2420 The RSSI value is always averaged over 8 symbols periods (128 us) and digitized by an 8 bit ADC The Link Quality Indication (LQI) measurements is a characterization of the strength and/or quality of received packet RSSI and LQI values are not necessarily linked; but Low LQI Invalid RSSI Chipcon CC2420 Radio The Chipcon CC2420 Radio Chip Characteristics Telos Mote Experiment, Stockholm, 2005 The Effect of Radio Interference The values of RSSI could really affected by other people working with motes Although there is no direct interference with the Wi-Fi channels, the channel quality can be affected by the overload and thus distort some of the packets The Effect of the Motes Themselves Measurements with the same motes at the same location, same batteries show a difference up to a few dBm (from new motes to old motes) Chipcon CC2420 Radio CC2420 Wireless Channel Characteristics RMS Delay Spread Delay Spread Td max i (t ) j (t ) , i, j : longestand shortest path i, j Td 30 m 1m 100 ns 3.3ns 97 ns 8 8 3 *10 3 *10 WC 1 5.7 MHz 2Td WSignal BW 2 MHz 5.7 MHz Channel is flat fading and a single channel filter tap is sufficient to model the channel RMS Delay Spread F=2.4 GHz 30 ns 1 50 660KHz WC 1 6.6 MHz 5 Corry Hall Corridor FCF 0.9 Frequency Selective FCF 0.5 Flat Fading Chipcon CC2420 Radio CC2420 Wireless Channel Characteristics Coherence Time TCohe 17.6ms v 3 m / s .423 C 3 108 .423 .423 fd vf c v 2.4 109 176ms v .3 m / s 137 TCohe NoRSSIS round RSSI 128 s MeasTime 1375 The maximum number of stable RSSI samples Coherence Dis tan ce CohTime *UserVeloci ty 5 cm 2 In our system, there is one RSSI sample per packet In this network configuration the RSSI rate is 26.3 sample/second per device CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) Rx Tx d1 d2 Run: RSSI_FigAvgVarHist.m at C:\ZigBeeNodes\CollectedData\Corridor_072106\Test1_QP CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) Worth Case Conclusions PathLoss vs Distance in Quiet Corridor 0 -17 dbm -20 Path Loss (dB) Theoritical PL Practical PL Real RSSI(d) = Pt(dBm) - PL(D0) - 10nlog10(d/D0) + Noise - Constant offset is not -45 dBm - Path Loss almost follows the LogNormal Model -40 RSSI(d) = -17.112 - 36.3236 - 1.85*10log10(d/3(ft)) -60 -80 -100 - The effect of multipath is remarkable RSSI(d) = -17.112 - 57.0773 - 1.69*10log10(d/3(ft)) 0 2 4 6 8 10 Distance (ft) 12 14 16 18 20 - Something is wrong in these sensors (Interference/Sensors Board) PathLoss vs Distance in the Presense of People at Corridor 0 -20 Path Loss (dB) Theoritical PL Practical PL Real RSSI(d) = Pt(dBm) - PL(D0) - 10nlog10(d/D0) + Noise -40 4d 0 PLd0 RSSI(d) = -17.112 - 34.9492 - 1.78*10log10(d/3(ft)) -60 -80 -100 RSSI(d) = -17.112 - 50.016 - 1.64*10log10(d/3(ft)) 0 2 4 6 8 10 Distance (ft) 12 14 16 18 20 n CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) RSSI Mean vs Distance In The Corridor 20 Max Err = 9.6015 & Mean Err = 3.1511 RSSI Mean 0 -20 Quiet Env People |Mean Err| -40 -60 -80 0 2 4 6 8 10 Distance (ft) 12 14 16 18 20 16 18 20 RSSI SD vs Distance In The Corridor 5 Quiet Env People RSSI Var 4 3 2 1 0 0 2 4 6 8 10 Distance (ft) 12 14 CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) Box-Whisker Plot, Env: Quiet Corridor, Black Circle: Mean Value -20 200 RSSI -30 RSSI (dBm) -40 -50 -60 -70 -80 1 2 3 4 5 6 7 8 9 10 11 12 Distance (ft) 13 14 15 16 17 18 19 20 21 CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) Box-Whisker Plot, Env: Not Quiet Corridor, Black Circle: Mean Value -20 200 RSSI -30 RSSI (dBm) -40 -50 -60 -70 -80 1 2 3 4 5 6 7 8 9 10 11 12 Distance (ft) 13 14 15 16 17 18 19 20 21 CC2420 RSSI Evaluation Path Loss Model Indoor Path Loss Model Test Scenario One sensor put at different distances (at the ~same time/Corridor) Variance Test Mica Z Min Var Max Var Our Sensors (Outdoor) 0.20 0.05 1.90 9.17 0.91 1.60 Our Sensors (Indoor) 0.02 0.20 2.17 19.20 2.40 7.68 Stanford Sensors 0.0 13.77 .02822 (Indoor & 100 Sensors ) Avg Var dBm Worth Case Conclusions -Something is wrong in these sensors - The effect of interference CC2420 RSSI Evaluation Path Loss Model Outdoor Path Loss Model Test Scenario Two sensors put at different distances (at the ~same time/Soccer Court) Conclusions RSSI Mean vs Distance In Large Outdoor Env 0 Device 1 Dev1-LogNrmPL Device 2 Dev2-LogNrmPL RSSI Mean (dB) -20 -40 - Constant offset is not -45 dBm - Path Loss almost well follows the Log-Normal Model -60 - In the outdoor these sensors have almost same behavior -80 -100 0 0.5 1 1.5 2 2.5 Distance (ft) 3 3.5 4 4.5 RSSI Mean vs Distance In Large Outdoor Env After Offset Compensation, Avg-PrcThrDiff ~ 2.5 dB 20 Device 1 Dev1-LogNrmPL Device 2 Dev2-LogNrmPL RSSI Mean (dB) 0 -20 -40 -60 -80 0 0.5 1 1.5 2 2.5 Distance (ft) 3 3.5 4 4.5 - There is some offset between sensors, so calibration is necessary CC2420 RSSI Evaluation Path Loss Model Outdoor Path Loss Model Test Scenario Two sensors put at different distances (at the ~same time/Soccer Court) Box-Whisker Plot, Dev 1, Env: OutDoor RSSI (dBm) -20 400 RSSI -40 -60 -80 1 2 3 4 Distance (ft) 5 6 7 Box-Whisker Plot, Dev 2, Env: OutDoor RSSI (dBm) -20 200 RSSI -40 -60 -80 1 2 3 4 Distance (ft) 5 6 7 CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI The Effect of Height Test Scenario Two different sets of data at two different height with 1.22 m height difference was compared (Test1 & Test2 was done at two different time/Env-Conditions) Height : 1.22 m height difference 6 dBm power difference Test 1 AvgErr AbsMeanErr 4.5767 dBm Test 2 AvgErr AbsMeanErr 7.4928 dBm Reported Error for the Same Chip is 10 dBm for 1m height difference S1: 4.6022 S2: 1.9371 S3: 0.0090 All in dBm S4: 11.2586 S5: 7.6377 S6: 2.0158 S1: S2: S3: S4: S5: S6: 4.1134 7.8446 12.3849 All in dBm 12.9487 0.4060 7.2592 CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI The Effect of Window Test Scenario Two different sets of data at two different situations (at the same place, time and Envcondition) was compared Window : Open/Closed 0.6 dBm power difference Test AbsMeanErr (All in dBm) S1: 1.9241 S2: 0.0140 S3: 0.9735 S4: 0.5581 S5: 0.0630 S6: 0.2506 Close to Window CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI The Effect of Door Test Scenario Two different sets of data at two different situations (at the same place, time and Envcondition) was compared Door : Open/Closed 1.5 dBm power difference Test AbsMeanErr S1: S2: S3: S4: S5: S6: (All in dBm) 3.5996 0.0005 0.7787 1.9850 1.9377 0.0354 Close to Door - Due to the need of calibration, error should be lower that this CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI The Effect of Lights Test Scenario Two different sets of data at two different situations (at the same place, time and Envcondition) was compared Lights : On/Off 1.7 dBm power difference Test AbsMeanErr S1: S2: S3: S4: S5: S6: (All in dBm) 7.7846 0.0319 1.0982 0.8703 0.5434 0.0461 In the order of several 0.1 dBm - Due to the need of calibration, error should be lower that this CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI The Effect of Antenna Direction Test Scenario Two different sets of data at two different antenna directions (at the same place, time and Env-condition) was compared Test AbsMeanErr (All in dBm) 5.5 dBm power difference - Due to the need of calibration, error should be lower that this Light Frame1 1 2 6 Mobile Node (Rx) 4 5 Light Frame2 Sensors (Tx) 3 ? Win Wal Brd Dor Radiated pattern Vertical mounting Radiated pattern horizontal mounting CC2420 RSSI Evaluation Feasibility Test Power Distribution Test Scenario Data Received by coordinator located under each sensor (at the time and Envcondition) was compared 100 50 0 10 5 2 4 100 - Due to the need of calibration, there is some uncertainty in these results. 50 0 10 5 100 50 2 0 0 Device 5 & Direction: Brd 4 100 -RSSI (dB) -RSSI (dB) 5 50 0 10 4 5 0 2 0 2 4 0 0 Device 2 & Direction: Brd -RSSI (dB) -RSSI (dB) 0 0 Device 4 & Direction: Brd 0 10 Conclusions Device 1 & Direction: Brd -RSSI (dB) -RSSI (dB) Device 3 & Direction: Brd 100 50 0 10 5 2 0 0 Device 6 & Direction: Brd 5 2 4 100 50 0 10 4 0 0 CC2420 RSSI Evaluation The Effect of Different Parameters on RSSI Basic Fingerprinting Radio Map Elements 48 runs in different situations (Quiet and Non-Quiet) [ Positioni (Cell Index) ( SensorID j , Direction k , TxPwl ) mean([ RSS Vector)] Static Test Expected Position 1 Estimated Position (SQR) 3 Estimated Position (ABS) 3 2 6 6 3 3 3 4 4 4 5 5 5 6 6 6 3 4 4 1 1 3 4 3 3 3 2 3 2 2 24 1 6 1 6 4 5 5 6 2 2 6 5 5 5 5 5 3 2 2 2 6 2 1 3 4 3 4 4 Max Error 2.44m MobileTest Expected Position: 3 4(2) 5(6) 6(5) 2 1 3 4(2) 5(6) 6(5) 2 1 Estimated Position: 4 2 4 2 4 6 4 6 4 6 4 6 3 6 3 2 2 2 2 2 2 5 3 2 3 6 1 2 3 2 1 2 2 2 5 2 5 2 Sensors By the offset compensation of sensors we can improve the performance Light Frame2 5 Mobile Node 6 4 3 2.44 m 1.65 m 2 1 Light Frame1 2 2 2 2 2 2 NI RSSI Evaluation NI Power samples n=2.2, AvgVar= 0.05 & MaxVar= 0.25 5 0 -5 Received Power (dBm) -10 Practical PL Theortical PL |Thr-Prc| -15 -20 -25 -30 -35 -40 -45 0 0.1 0.2 0.3 0.4 0.5 Distance (m) 0.6 0.7 0.8 0.9 Thank You The End Box Whisker Plot (N(0,sigma))