1 PinPoint: An Asynchronous Time Based Location Determination System Moustafa Youssef, Adel Youssef, Chuck Rieger, UdayaShankar, Ashok Agrawala Presented by Sofia Nikitaki
2 Motivation Location-aware applications Software-based Standard protocols Rapidly deployable No calibration Energy-efficient Number of messages
3 Review Motivation Related Work PinPoint Technology PinPoint Hardware Results Conclusions
4 Time-based Localization Problem Clock synchronization Solutions 1.Use synchronized clocks: GPS Expensive 2.Use echoing (round trip time measured by same clock) Time measurement is not precise Affected by processing time 3.Use two different signals: Cricket RF for synchronization Ultrasound for ranging 4.PinPint
5 The PinPoint Technology Time of arrival based Software only solution: can work with standard protocols (WiFi, WiMax) Efficient (constant number of messages per node to locate all other nodes Works both indoors and outdoors
6 PinPoint Technology 3 Phases Measurement phase Information Exchange phase Computation Phase
7 PinPoint Technology: Measurement Phase node message (ID, transmit) records the receive timestamp (the messages sent by other nodes) All messages are one way and broadcast Inter-node distances, hence spatial layout (from trilateration) Inter-node clock drifts and offsets, hence ability to carry out synchronous action with other nodes
8 Clock Model Node’s clock is assumed to have drift stable over short periods. Clock time τ is related to the real time t by Where: a, b constant, measurement phase. B drift rate (no worse than 100 parts per million) t is measured with a nanosecond resolution
9 PinPoint TechnologyGlobal O(n) messages for n nodes Let ta1, tb1: tx, rx ts of first A msg tb2, ta2: tx, rx ts of first B msg ta3, tb3: tx, rx ts of second A msg ta4, tb4: tx, rx ts of second B msg
10 PinPoint Technology: Information Exchange Phase node message containing its receive timestamp for messages transmitted by other nodes O(n) messages
11 PinPoint Technology: Computation Phase node computes spatial coordinates clock attributes of every other node Redundant information used to reduce errors No communication takes place Accuracy: few feet Synchronized clocks
12 Calculations for Node Pair A and B Drift ratio Propagation delay Remote clock reading
13 Youssef14PinPoint Estimators Clock model Each node has its offset (a)and drift rate (b) from the global time t= b(a+ t) local_time= drift_rate* (offset+ global_time) Distance estimate: b b d B b is in the order of hundred parts in a million
14 PinPoint Hardware
15 PinPoint Hardware Altera Cyclone 1C20 FPGA development kit computation processor and 3 ns timestamping clock Maxim 2820 radio with Maxim 2242 RF power amp communication modules for transmitting the signal used for timestamping MaxStream 9xStream radio modem Information exchange
16 PP2: Timestamping 2.4 GHz QPSK modulation PP2 baseband signal Received signal processed as follows Zero-crossings detected and time stamped Many more zero-crossings detected than transmitted (noise, multi- path) Choose “longest chain” of zero-crossings as true signal Do least-squares fit to obtain arrival time of signal Works well indoors (detects first signal in multi-path)
17 Distance Vs Reported Clock Ticks 4-6 feet accuracy using 3 ns clock
18 Indoors Evaluation-First Testbed 4.18 feet average error 8.42 feet at 95% 37 feet range
19 Indoors Evaluation-Second Testbed All locations NLOS 4.95 feet average error 11 feet at 95% 74 feet range
20 Outdoors Evaluation 6.85 feet average error feet at 95% 146 feet range
21 Mobility Evaluation Can track the user in realtime PinPoint exchange cycle is much faster than user mobility rat
22 Conclutions A time-based ranging technology asynchronous clocks no echoing constant number of messages per node Can synchronize clocks Can work with standard protocols Does not require calibration 4-6 feet accuracy ( m)
23 Thank You