An E-Toll Transponder Network for Smart Cities Omid Abari Deepak Vasisht, Dina Katabi, Anantha Chandrakasan Electronic Toll Transponders 80% of the cars in US have them Some states are making it mandatory Opportunities 7 6 5 4 1 2 3 Opportunities Adjust traffic-light timing Smart Detect red-light runner street-parking One infrastructure for many smart services Challenge: No MAC to Prevent Collision Wireless query One car responds Physical Isolation Wireless query Interference All cars respond What can we do? at the same time I don't care 'Cause I've never been so high Follow me to the dark Let me take you past our satellites Caraoke • A system that delivers smart services using existing e-toll transponders despite interference • Software-Hardware solution: – Count cars – Localize cars – Decode transponders • Built and evaluated on our campus streets with existing e-toll transponders Count cars: How to count despite interference? Localize cars Decode transponders Counting Despite Interference Freq-Domain Time-Domain 𝛿 𝑓 − 𝑓𝑐 + 𝑆 𝑓 − 𝑓𝑐 Data Spike Amplitude · 𝑒 𝑗2𝜋𝑓𝑐 𝑡 Carrier Frequency Counting Despite Interference Freq-Domain Time-Domain 𝛿 𝑓 − 𝑓𝑐 + 𝑆 𝑓 − 𝑓𝑐 Data Spike Amplitude · 𝑒 𝑗2𝜋𝑓𝑐 𝑡 Carrier Frequency Amplitude Toll transponders have difference in their carrier frequency Caraoke can count transponders despite interference Frequency Count cars Localize cars: How to localize despite interference? Decode transponders Localizing Transponders Localization requires knowing the channel How to extract channels in the presence of interference? Localizing Transponder Time-Domain · 𝑒 𝑗2𝜋𝑓𝑐 𝑡 ℎ1 · Freq-Domain ℎ1 · 𝛿 𝑓 − 𝑓𝑐 +ℎ1 𝑆 𝑓 − 𝑓𝑐 Spike ℎ1 ℎ3 ℎ2 ℎ1 ℎ2 ℎ1 ℎ3 Frequency Use the channel to find the Angle-of-Arrival Localizing Transponder z y x x Localizing Transponder Cars are always on the road plane z y x Intersect spatial angle with road plane Hyperbola Intersects hyperbolas across multiple readers to find the exact location of the car Count cars Localize cars Decode transponders: How to decode despite interference? Imag Solution: Coherent Combining Real Imag Solution: Coherent Combining Real Imag Solution: Coherent Combining Real + + = Caraoke uses channel information to combine responses coherently and decode the ID Custom Hardware Design • Self-Sustaining – Harvest Solar Energy Custom Hardware Design • Self-Sustaining – Harvest Solar Energy • Low-Cost – Replace TX chain with simple PLL Power: 9mW; Cost: $40 Experimental Results Evaluation • MIT campus- four streets • Caraoke readers were placed on 12.5-feet poles • Standard E-ZPass transponders on the cars Accuracy of Counting Transponders Accuracy (%) 100 98 96 94 92 90 10 20 30 40 Number of Transponders 50 Higher accuracy and significantly lower cost than camera-based systems [R.L., JEI’13; J.M. Urbana‘08 ] Accuracy of Localizing Transponders 1 CDF 0.8 0.6 0.4 0.2 0 0 2 4 6 8 Localization Error (in degree) Caraoke enables enough accuracy to detect occupied versus available parking spots Detected Speed(mph) Speed Detection Accuracy 60 50 40 30 20 10 0 10 20 30 40 50 Actual Speed(mph) Caraoke detects the speed to within 8% Identification Time (mS) Required Time for Decoding 60 50 40 30 20 10 0 1 2 3 4 5 6 7 8 9 10 Number of Transponders Caraoke enables real-time smart city services Related Work • Wireless Interference – They require changing transmitter or leverage intrinsic asynchrony in protocols (S. G. Sigcomm’12, J. W. Sigcomm’08, etc.) • Smart Parking – They require sensors on parking spots or Wi-Fi devices in cars (Street Line, S. M. MobiSys’10, S. N. Mobicom’13, etc.) • Automated Payment – They require a gate and physical isolation (McDonald’s drive-through) Conclusion • A system for delivering smart services using existing e-toll transponders • Can count, localize and decode transponders in the presence of interference • We built it into a small PCB An E-Toll Transponder Network for Smart Cities Omid Abari Deepak Vasisht, Dina Katabi, Anantha Chandrakasan
© Copyright 2024 Paperzz
