Frequency Hopping

advertisement
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Polly Huang
Electrical Engineering
National Taiwan University
http://nslab.ee.ntu.edu.tw/
1
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Road Map
 Sensor network?
 Indoor localization?
 The system & testbed
 Technical challenges
 Causality Study
 Frequency Hopping
 Hospital Installation
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Sensors?
camera
pressure

mic
accelerometer
thermal
gyro
GPS
Also the biomedical sensors: EMG, EKG, pulses,
emotions, etc
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Sensor Nodes Today
MICA, 2001-2002
5.7cm X 3.18cm
Spec, March 2003
2mm X 2.5mm
4 MHz CPU
128K ROM 512K RAM
40kbps Radio range x00 feet
Sensors, battery not included
CPU, memory, RF transceiver
Sensors, battery, antenna not
included
< 1 dollar if mass-produced
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Embedded Sensor Node
Intel® Xscale CPU
Analog and digital radio
Flash and SRAM memory
Sensors
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Embedded Sensor Network
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Localization Systems
 Outdoor
 GPS
 Indoor
 UWB, ultrasound,
infrared, etc
 RF-based system
9
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Radio as Location Sensor
 802.15.4 (Zigbee) radio
 RF radio
 Active RFID
Beacon
Tag
Signal Strength ~ Distance
Propagation Time
Photo courtesy: Crossbow
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Classic Techniques
 Ranging & Triangulation
 Each radio source to a receiver
 RSSI ~ distance
 Radio Fingerprinting
 A bunch of radio sources to a receiver
 RSSIs(position_X) ≠ RSSIs(position_Y)
11
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Localization - Triangulation
(x1, y1, z1)
(x2, y2, z2)
Infrastructural
Beacons
Mobile
Users
(x3, y3, z3)
(x5, y5, z5)
(x4, y4, z4)
d1
d2
(x, y, z)
d3
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Requirement
 Important property of the electromagnetic wave used
 Radio time of flight transmitter  receiver
 Proportional to distance transmitter  receiver
 High frequency EM waves, e.g.,
 Infrared
 Ultrasound
 Problem
 Line of sight
 Placement of tags a challenge
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Localization – Radio Fingerprint
Fingerprint location
Beacon 1
Beacon 2
Beacon 3
(B1,B2,B3,B4…) (x1,y1,z1)
…..
…..
(B1,B2,B3,B4)
…..
…...
Beacon 4
Look up the
table
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Road Map
 Sensor network?
 Indoor localization?
 The system & testbed
 Technical challenges
 Causality Study
 Frequency Hopping
 Hospital Installation
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
1. Survey Phase
Radio Fingerprints
Location A
Beacon 1 = -65 Beacon 2 = -70 …
Location B
Beacon 1 = -70 Beacon 2 = -78 …
Location C
Beacon 1 = -71 Beacon 2 = -85 …
Location D
Beacon 1 = -72 Beacon 2 = -77 …
…
…
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
2. Tracking Phase
Radio Fingerprints
Location A
Beacon 1 = -65 Beacon 2 = -70 …
Location B
Beacon 1 = -70 Beacon 2 = -78 …
Location C
Beacon 1 = -71 Beacon 2 = -85 …
Location D
Beacon 1 = -72 Beacon 2 = -77 …
…
…
RSSI
Location ?
Beacon 1 = -65 Beacon 2 = -70 …
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Location Estimation
 K-Nearest Neighbor(KNN)
 K=3
Tracking :
Beacon 1=-70 Beacon 2=-77 …
Radio Fingerprints
Location A
Beacon 1 = -65 Beacon 2 = -70 …
Location B
Beacon 1 = -70 Beacon 2 = -78 …
Location C
Beacon 1 = -71 Beacon 2 = -85 …
Location D
Beacon 1 = -72 Beacon 2 = -77 …
…
…
K closest fingerprints
(with minimum
Euclidean distance)
Location B @(x_B,y_B)
Location D @(x_D,y_D)
Location F @(x_F,y_F)
Output
Position
Weighted Average
18
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Testbed Deployment
Training area
USB Hub
PC gateway + Power supply
Beacon nodes
USB Extender
Modulator
Demodulator
19
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Maximum Transmission Distance:
Issue I
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Power Supply: Issue II
Tail of the chain
Head of the chain
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Power Confusion: Issue III
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Localization Error
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Road Map
 Sensor network?
 Indoor localization?
 The system & testbed
 Technical challenges
 Causality Study
 Frequency Hopping
 Hospital Installation
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Antenna Orientation: Experiment I
Average location
error:
0.88 m
Average location
error:
2.05 m
Average location
error:
3.17 m
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Antenna Radiation Pattern
Radiation Pattern
On-board
antenna
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Foreground Obstacle: Experiment II
Tag on chair
Tag in hand
Human standing still
Human wandering around
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Foreground Obstacle Experimental Result
• RSSI of tag on chair: More stable
• RSSI of tag held in hand: High in variance
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Foreground Obstacle Experimental Result
(a)
• RSSI of tag on chair: More stable
• RSSI of tag held in hand: High in variance
(b)
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Foreground Obstacle Experimental Result
(a)
• RSSI of tag on chair: More stable
• RSSI of tag held in hand: High in variance
(b)
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Background Noises: Experiment III
Daytime
• Average localization error:
• 12:00AM – 7:00AM: 1.58m
• 9:00 AM – 12:00AM: 1.52m
Midnight
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Beacon Density: Experiment IV
X1 X2
1/3 beacons (I)
X1 X2
1/3 beacons (II)
X1 X2
1/3 beacons (lll)
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Take Aways
• Major factors:
• Antenna orientation
• Foreground obstacles
• Suggestions:
• Use of uniform radiation pattern antenna
• Same antenna orientation in training and tracking
• The method in training and tracking should be
•
consistent.
Denser beacon deployment  higher accuracy
• Not an issue:
• Background noise
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Road Map
 Sensor network?
 Indoor localization?
 The system & testbed
 Technical challenges
 Causality Study
 Frequency Hopping
 Hospital Installation
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Radio Interference
A Common Problem
Beacon node 1
(beacon1, RSSI1)
(beacon2, RSSI2)
(beacon3, RSSI3)
(beacon4, RSSI4)
Location estimation

Beacon node 2
.
Beacon node4
Beacon node 3
Interference
35
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Interference  Localization Error
 Implementation
 Localization


Zigbee
RSSI-signature
 Data

WiFi
 Experiments
 WiFi data rate
 68 – 2835 kbps
1.60m
3.85m
36
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Cause: Beacon Packet Loss
37
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Frequency hopping to mitigate the impact of
interference on localization accuracy!
Longer beacon packet collection time?
1. Location sampling delay
2. Data traffic drop due to interference
38
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Contributions
 Quantify the impact of interference
Done!
 Devise a frequency hopping mechanism
Next!
 Evaluate localization stability under realistic
influences
39
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Mechanism Components
 How to hop
 In a relay fashion
 When to hop
 A diagnostic test decide whether to hop
40
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Hopping Notification
Diagnostic
Test
Hop
Timeout
41
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Diagnostic Test – 3 Parts
 Easy part
 Not-so-easy part
 Difficult part
42
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
The Easy Part
 Strong evidence
 Beacon loss ↑  Location error ↑
 Beacon packet loss
 A good criteria
 Packet Reception Rate (PRR)
 PRR
 # packets received in T / # packets sent in T
 5 packets/sec
43
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
The Semi-Easy Part
 Beacon loss ↑  Location error ↑
 Observed at the mobile tag
 Need to know
 PRR from a beacon when there’s no interference
 So mobile tag can tell if PRR is relatively ↓
 PRR from a beacon
 Differ depending on Location of mobile tag
44
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
The Solution
 Every beacon node listens to other beacon nodes!
 At system setup (no interference)
 Recording PRR from all other beacons
 Select good neighbor beacons
 During operation
 Record PRR from good neighbor beacons
 Use this PRR to determine whether there’s interference
45
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
The Difficult Part
 How bad PRR  interference
 Too low => beacon packet losses
 Too high => hop unnecessarily
46
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Proposed Solutions
1. Thresholding
 Measure the system for a while
 Find a proper threshold that minimize error
 Brute-force search (naïve >”<)
2. Hidden Markov Model(HMM)
 Measure the system for a while
 Train a model based on measured relationship between

PRR vs. localization errors
 More scalable (scientifically sound)
47
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Why HMM?
 Time dependency in background WiFi traffic
 Quiet  Quiet & Busy  Busy (Often)
 Quiet  Busy & Busy  Quiet (Seldom)
transit ↓
Quiet
Busy
remain ↑
 Time dependency in PRR
 Background WiFi Traffic ↑  Beacon PRR↓
 HMM models temporal pattern in PRR
 Transition probability
48
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Data Sets: Real WiFi Usage
49
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Localization Errors
1.32m
1.82m
1.24m
2.74m
a slight penalty of receiving
no beacon packets during
transition of hopping
50
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Hopping delay
 Loss pattern of hopping message ~ beacon message
 Time for all beacons to hop to the next channel is
simulated by beacon PRR
No Timeout
51
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Take Aways
 RSSI-based localization system sensitive to
interference
 Localization error under interference
 Degrade by 141%
 Frequency hopping effective
 Localization error with frequency hopping
 Reduced to 45%
52
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Road Map
 Sensor network?
 Indoor localization?
 The system & testbed
 Technical challenges
 Causality Study
 Frequency Hopping
 Hospital Installation
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
54
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
室內定位系統
 甚麼是定位系統?
 找出人或物品目前所在的位置
 市面上有沒有商品化的定位系統
 GPS衛星定位

可是,衛星定位只能用在室外
 室內定位系統
 目前有幾家,使用的無線技術包括WiFi,UWB、主動式RFID等
 而我們的系統使用的是Zigbee
 室內定位系統長期持續使用於醫護場所中
 就我們所了解,目前北護第一家
55
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
室內定位系統有甚麼用?
 以北護來說
 可經由住民目前所在位置,了解其意圖以提供協助,帮
助維護住民安全及減少醫護人員負擔。
 例如:住民意圖自行離開護理之家、住民進入廁所等
 在其他場所
 智慧型居家
 室內導航/導覽:博物館、展覽館、百貨公司等
 安全維護
56
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
定位系統長甚麼樣?
57
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
佈建在天花板
裡面的”衛星”
定位系統組件
定位系統所用
的追蹤器
追踨畫面顯示
58
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
佩戴方式
59
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
開啟系統 – 步驟(1)
在定位系統圖
示上點兩下
60
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
警戒區
目前共設定5個警戒區
61
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
警示: 進入警戒區
1. 跳出警告視窗
2. 發出警告音樂
3. 使用者追踨點的
顏色會改變
62
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
麻煩護理人員前往處理
系統發出警告時,麻煩
護理人員前往處理,然
後回報系統
住民確實進入警
示區,請點選”
完成“
住民沒有進入警示區,
系統卻發出警示,請
點選“系統誤報”
63
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
電池更換充電
 每位使用者都準備兩個一模一樣的追踨器
 可交互替換使用
 更換時間
 每隔兩天更換一次
 目前暫定星期一、星期三、星期五、星期日
64
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
更換方法-1
替換追踨器的編號
必需與原本的相同
往下為關閉
步驟一:
收回使用者配帶的追
踨器,關上電源開關
(開關往下)
步驟二:
將充電座上已充好電的替換追踨器取下,然
後把已使用過的追踨器接到充電座上。請注
意, 兩個相互替換的充電器的編號必需相同
65
Network and Systems Laboratory
nslab.ee.ntu.edu.tw
Polly Huang
Electrical Engineering
National Taiwan University
http://nslab.ee.ntu.edu.tw/
66
Download