Emerging Technologies in Smart Device Communications Mary Ann Ingram School of Electrical and Computer Engineering Georgia Institute of Technology TR-50 Meeting April 12, 2010

Overview   Objective   Cooperative Transmission   Energy Harvesting and Storage   RFID and Sensors   Wake-up Radios   Conclusions   How Georgia Tech can help

Objective   Consider some emerging technologies that would impact the standard, because they impact the Data Link Layer (DLL)   Identify some DLL issues for each technology

Cooperative Transmission (CT) Overview   Definition and SNR advantage   Transmit time synchronization   Range extension   Application to the energy hole in wireless sensor networks   Application to broadcasts in dense networks   DLL issues

Definition of CT   A protocol where multiple, neighboring radio platforms cooperate in the physical layer to send a single message I’ll help you if you will help me! OK! That way, if your channel is faded, maybe mine will be better! Working together, our signals can go farther and we may save energy! J. Laneman, D. Tse, and G. Wornell, “Cooperative diversity in wireless networks: Efficient protocols and outage behavior,” IEEE Trans. on Information Theory, vol. 50, no. 12, pp. 3062–3080, Dec

CT Gives an SNR Advantage   Diversity gain – – >13 dB in Rayleigh-fading channels   Array gain (e.g. 6 dB with 4 nodes) – –When all nodes transmit with same power   Extra SNR can be used for range extension, TX power reduction, lower PER, higher data rate

Concurrent CT (CCT)   Cooperating nodes transmit at the same time – –A commonly received packet provides reference for synchronization   The alternative is time-division CT – –Nodes transmit in non-overlapping time periods   CCT is better for range extension – –Receivers get the benefit of diversity gain for synchronization – –Less SNR loss from imperfect RX synchronization

Concurrent CT (CCT) Transmit Time Synchronization   Non-coherent FSK   Median rms TX time spread for -5 dBm TX pwr – –First hop 60 ns – –Other hops ~125 ns   Will see in the demo Chang and Ingram, "Convergence Property of Transmit Time Pre-Synchronization for Concurrent Cooperative Communication," submitted to Globecom 2010

CCT Range Extension – Dispersed Cluster   Light grey = 2-hop non-CT coverage   Light + dark grey = 2-hop CT coverage – –80% increase Jung, Chang and Ingram, "Comparison of Two Cluster Topologies for Cooperative Transmission Range Extension in the 2.4GHz Band," submitted to Globecom 2010

CCT Range Extension – Tight Cluster   CT yields 280% area increase   But total 2-hop dispersed CT area is 55% larger than tight Jung, Chang and Ingram, "Comparison of Two Cluster Topologies for Cooperative Transmission Range Extension in the 2.4GHz Band," submitted to Globecom 2010

The Energy Hole Problem in Battery-Driven WSNs  The nodes near the sink have to relay the data from the rest of the network, and die early Sink A single sink

Conventional Approaches to Mitigate the Energy Hole  Non-uniform distribution [Wu08] –Placing more nodes in the area close to the sink –The extra sensors can significantly raise the cost  Using mobile sensors [Wang05] –May not be possible depending on the environment and the hardware [Wu08] X. Wu, G. Chen, and S. K. Das, “Avoiding energy holes in wireless sensor networks with nonuniform node distribution,” IEEE Trans. Parallel Distrib. Syst., vol. 19, no. 5, pp. 710–720, [Wang05] W. Wang, V. Srinivasan, and K.-C. Chua, “Using mobile relays to prolong the lifetime of wireless sensor networks,” in Proc. IEEE MobiCom, 2005.

Using CT to Extend Network Life  Extend range with CT to jump over heavily-loaded nodes  Simulation: Factor of 8X lifetime extension Conventional (Non-CT) Routing Cooperative Routing Non-CT Flow Virtual MISO Link Jung and Ingram, "Residual-Energy-Activated Cooperative Transmission (REACT) to Avoid the Energy Hole," ICC CoCoNet Wkshp, 2010.

Opportunistic Large Array   A group of nodes that, without coordinating with each other, transmit the same message at approximately the same time in response to a signal received from another transmitter or OLA   Uses: – –Fast, contention-free, reliable broadcasts – –Complexity is independent of density for high- density networks – –Simple OLA-based unicasting is available * A. Scaglione, and Y. W. Hong, IEEE Trans.Signal Processing, 2003.

OLA 1 Decoding Level 1 (DL 1 ) DL 1 DL 2 DL 3 OLA 2 Decoding Level 2 (DL 2 ) Decoding Level 3 (DL 3 ) In more energy efficient versions, only subsets transmit OLA Broadcasting * A. Scaglione, and Y. W. Hong, IEEE Trans.Signal Processing, Faster than multi-hop because no contention

Some DLL Issues   CT is not supported by current protocols   Cooperators have to be selected – –Can be done autonomously in sufficiently dense networks   CCT requires a commonly received packet to provide a synchronization reference – –Need a field in the header to command the CCT   CCT node transmissions must not include their addresses – –Received signal must appear to have come from a single node through a multi-path channel   Distributed ARQ required for OLA transmissions (OLA has no cluster heads)   For energy balancing, the set of all potential cooperators must be informed of Sink Nav

Overview   Objective   Cooperative Transmission   Energy Harvesting and Storage   RFID and Sensors   Wake-up Radios   Conclusions   How Georgia Tech can help

Energy Harvesting and Storage   Success of embedded/pervasive devices depends on success of energy harvesting   Device technologies are developing fast – –Harvester prototypes for almost every kind of energy – –Amounts of energy harvested differ by orders of magnitude (solar highest; RF lowest) – –Storage technologies are also developing fast   MAC/routing research is slowed because of inadequate system-level models for harvester and storage M.A. Ingram et al., “Energy Harvesting Wireless Sensor Networks,” in Globalisation of Mobile and Wireless Communications: Today and in 2020, R. Prasad, ed., Springer, to appear Spring 2009.

Two Energy Storage Options - Opposites in Many Respects   Rechargeable battery (RB) – –High energy density – –Low peak power – –Low leakage – –Few hundred recharge cycles – –Constant voltage – –Sensitive to depth of discharge 19  Supercapacitor (SC) –Low energy density –High peak power –High leakage –Million recharge cycles –Variable voltage –Not sensitive to depth of discharge

Harvester Power Matching   Harvester has low efficiency if storage device is not impedance matched to source   Source impedance varies   Optimal matching circuits have to track the source, but can consume too much energy themselves

DLL Issues   Harvesting takes time- – –Devices can be unavailable for minutes to hours while harvesting   Duty cycling becomes problematic because of random node availability – –Adaptive control theory proposed for aperiodic energy sources

Overview   Objective   Cooperative Transmission   Energy Harvesting and Storage   RFID and Sensors   Wake-up Radios   Conclusions   How Georgia Tech can help

RF Tag Review   Mature technology   Optical bar code replacement   Passive RF Tag – –No battery on board – –Processor activated by a reader’s RF signal – –Reader has to be close (few meters) – limited by activation – –Modulated backscatter for transmission   Eliminates power amplifier   Semi-passive RF Tag – –Battery on board, runs the processor – –Modulated backscatter for transmission – –Read range longer- limited by mod backscatter

New: RF Tag + Sensor + Energy Harvesting   Two-tier network – –Top tier: mesh network of readers   Access power mains – –Bottom tier: low-power semi-passive sensors   Powered by harvested ambient energy   Communicate with top tier by modulated backscatter   Dedicated source can provide RF for energy harvesting (PowerCast) – –This source need not be a communicating node – –Powers nodes where there is no ambient energy   Behind walls, low-light areas, above the ceiling Clark et al., "Towards Autonomously-Powered CRFIDs," Workshop on Power Aware Computing and Systems (Hot-Power ’09), October 2009 A. Sample and J. R. Smith, "Experimental Results with two Wireless Power Transfer Systems," 2009.

DLL Issues   MAC needed for reader transmissions   Multiple readers can collide trying to read the same tag   Reader TX power for modulated backscatter is higher than traditional radio- larger interference range   Must provide time for power-up delay Waldrop et al, “Colorwave: A MAC for RFID Reader Networks,” 2003 G. P. Joshi, S.W. Kim, “Survey, Nomenclature and Comparison of Reader Anti-collision Protocols in RFID,” IETE Tech Review [serial online’ 2008.

Overview   Objective   Cooperative Transmission   Energy Harvesting and Storage   RFID and Sensors   Wake-up Radios   Conclusions   How Georgia Tech can help

Wake-Up Radios   A well known significant source of energy drainage is radio idle listening   Traditional power management uses duty cycling – –Nodes periodically wake up to check if they are needed. Most of the time they are not needed.   An ultra low-power radio can be used to trigger a remote interrupt at the sleeping device when communication with the device is required   Enables more efficient utilization for event-based and on-demand applications Van der Doorn et al, “A prototype low-cost wakeup radio for the 868 MHz band,” Int. J. Sensor Networks, Vol. 5, No. 1, 2009

Some Example Wakeup Radio Numbers   ~20 W to 170 W consumption   -75 dBm 915 MHz   0.5 m to 2 m range from 0 dBm transmitter Le-Huy and Roy, “Low-Power Wake-Up Radio for Wireless Sensor Networks,” Mobile Networks and Applications, April Van der Doorn et al, “A prototype low-cost wakeup radio for the 868 MHz band,” Int. J. Sensor Networks, Vol. 5, No. 1, 2009

Comparison to RFID   Wake-up radio operates like passive RFID   Wake-up radio is less complex than RFID and requires less power to be energized

Comparison to “Wake-On Radio”   In wake-on radio, the radio wakes up periodically to listen for incoming packets without microcontroller interaction   Wake-up radio does not wake up periodically Lu et al, “A Wake-On Sensor Network,” Sensys, 2009

DLL Issues   How often should wake up signals be sent?   Wake up signal wakes all nodes in the neighborhood   Collisions can happen between nodes sending wake up signals

Conclusions   CT offers many advantages for networks of highly energy-constrained radios; CT is not supported by current protocols   Energy harvesting and storage makes duty- cycling difficult – good models lacking   Wake-up radios can make duty cycling much easier; current technology is short-range   RFID with energy harvesting is sustainable, but needs reader   All these technologies would significantly impact the standard

How Georgia Tech Can Help With Standards Development   Objective technology assessment   Determine protocol changes to support specific technologies   Prototype development and prototype testing   Channel and device modeling   Certification test design

Thank You!