Wireless Sensor Networks Semiannual update Kris Pister Professor

Wireless Sensor Networks Semi-annual update Kris Pister Professor EECS, UC Berkeley (Founder & CTO, Dust Networks)

Outline Standards Technology Products & Applications 1

Outline Standards – HART 7, Wireless. HART – SP 100 – IETF – 6 Lo. WPAN, RSN – IEEE – 802. 15. 4 E – Zigbee Technology Products & Applications 2

Wireless. HART – Highway Addressable Remote Transducer – Wired standard, ~2 decades old – 25 million sensors deployed – 2 -3 million ship per year Industrial Standard – 54% of “field devices” (sensors) – >> $1, 000 for installation average Wireless HART – Part of HART 7 release – Ratified Sept 7, 2007 3

The De-facto Standard – October 2006 11 manufacturers, 95% of market, all talking w/ pre-HART TSMP Emerson MACTek Yokogawa Siemens Phoenix Contact Endress+ Hauser Elpro ABB Honeywell Smar Pepperl+ Fuchs 4

5

ISA SP 100 Principles of Operation published – Open conflict – Repeat of Fieldbus wars Wireless HART compatibility – – – Same radio PHY Same Network and DLL/MAC (TSMP) Same applications Same crypto Same customers But… – Egos & Fortunes Users furious – will it matter? 6

IETF Internet Engineering Task Force – Folks who standardized IP, TCP, HTTP, … 6 Lo. WPAN – IPv 6 over Low power Wireless Personal Area Networks – Draft standard submitted – UCB’s David Culler, Arch Rock heavily involved – Describes header compression for IPv 6, TCP, UDP • “only 2 bytes!” RSN – routing in sensor networks – “ok, so now what? ” Great start, good community Likely to be the death of Zigbee 7

IEEE 802. 15. 4 E Working Group From the PAR: “The intention of this amendment is to enhance and add functionality to the 802. 15. 4 -2006 MAC to – better support the industrial markets” – explicitly calls out HART and SP 100 elsewhere “Specifically, the MAC enhancements to be considered will be limited to the following: • TDMA: to provide a) determinism, b) enhanced utilization of bandwidth • Channel Hopping: to provide additional robustness in high interfering environments and enhance coexistence with other wireless networks • GTS: to increase its flexibility such as a) supporting peer to peer, b)the length of the slot, and c) number of slots • CSMA: to improve throughput and reduce energy consumption • Security: to add support for additional options such as asymmetrical keys • Low latency: to reduce end to end delivery time such as needed for control applications” 8

TSMP Time Synchronized Mesh Protocol Basis for – Wireless HART – ISA SP 100 – 802. 15. 4 E? – IETF/RSN? ? 9

Low Data Rate WPAN Applications Zigbee Pro 2004 2006 security HVAC AMR lighting control access control asset mgt process control environmental energy mgt BUILDING AUTOMATION CONSUMER ELECTRONICS PC & PERIPHERALS INDUSTRIAL CONTROL patient monitoring fitness monitoring TV VCR DVD/CD remote PERSONAL HEALTH CARE RESIDENTIAL/ LIGHT COMMERCIAL CONTROL mouse keyboard joystick security HVAC lighting control access control lawn & garden irrigation 10

(… but I heard that it doesn’t work …) Industrial automation – Decades of false starts Defense – NEST fallout Home, health care, HVAC, security, … – Failed investments in using chips & stacks – “We tried it w/ x, and couldn’t get it to work” 11

How reliable do you need your network to be? ? 12

How reliable do you need your network to be? ? 13

Outline Standards Technology – TSMP Products & Applications 14

Barriers to Adoption On. World, 2005 15

TSMP Foundations Time Synchronization – Reliability – Power – Sensor time stamps Reliability – Frequency diversity • Multi-path fading, interference – Spatial diversity • True mesh (multiple paths at each hop) – Temporal diversity • Secure link-layer ACK Power – Turning radios off is easy – Knowing when to turn them back on is tricky 16

Power-optimal communication Assume all motes share a network-wide synchronized sense of time, accurate to ~1 ms For an optimally efficient network, mote A will only be awake when mote B needs to talk A A wakes up and listens B B transmits B receives ACK A transmits ACK Expected packet start time Worst case A/B clock skew 17

Packet transmission and acknowledgement Mote Current Radio TX startup Packet TX Radio TX/RX turnaround ACK RX Energy cost (2003): 295 u. C 18

Timing – imperfect synchronization (latest possible transmitter) A B CCA: RX startup, listen, RX->TX RX startup Tg Tg Transmit Packet: Preamble, SS, Headers, Payload, MIC, CRC RX packet Tcrypto Verify CRC RX startup or Tx->Rx Verify MAC MIC Tg ACK RX ACK Calculate ACK MIC+CRC Transmit ACK RX->TX Expected first bit of preamble TCCA = 0. 512 ms to be standards compliant – Worst case is a receive slot followed by a transmit slot to a different partner, as radio will be finishing up the ACK TX just as it needs to look for a clear channel, so – TCCA = TTX->RX + Tchannel assessment + TRX->TX = 0. 192 ms + 0. 128 ms + 0. 192 ms Tpacket = 4. 256 ms for a maximum length packet – Preamble+SS+packet = 4+1+128 B = 133 B = 1064 bits 4. 256 ms @ 250 kbps Tcrypto is the total time to verify packet MIC and create ACK MIC Tg. ACK is the tolerance to variation in Tcrypto and/or mote B’s turnaround time from RX to TX TACK is a function of the ACK length. It is likely to be just under 1 ms. Tslot = TCCA+2*Tg+Tpacket+Tcrypto+Tg. ACK+TACK = 0. 512+2+4. 256+1+0. 1+1 = 9 ms 19

Fundamental platform-specific energy requirements Packet energy & packet rate determine power – (QTX + QRX )/ Tcycle – E. g. (60 u. C + 40 u. C) /10 s = 10 u. A 20

Idle listen (no packet exchanged) Mote Current Radio RX startup ACK RX Energy cost (2003): 70 u. C 21

Scheduled Communication Slots Mote A can listen more often than mote B transmits Since both are time synchronized, a different radio frequency can be used at each wakeup Time sync information transmitted in both directions with every packet A B TX, A ACK B Ch 3 Ch 4 Ch 5 Ch 6 Ch 7 Ch 8 QTX , QRX , Qlisten 22

Latency reduction Energy cost of latency reduction is easy to calculate: – Qlisten / Tlisten – E. g. 20 u. C/1 s = 20 u. A Low-cost “virtual on” capability Latency vs. power tradeoff can vary by mote, time of day, recent traffic, etc. 23

Multi-hop routing Global time synchronization allows sequential ordering of links in a “superframe” Measured average latency over many hops is Tframe/2 G T 2, ch y A T 1, ch x B Superframe 24

Link 53 37 A link defines a timeslot and channel offset. A link consumes capacity. Links are directional. Links are associated with graphs. This is the link that uses the blue graph connection from 37 to 53 on slot 4 and offset 7 25

Link types one destination > one • Contention free one source • Collisions possible > one • Unicast • Acknowledged • Broadcast • No ACK 26

Graph 17 47 53 37 41 43 A graph is a collection of connections and the links belonging to them A graph contains both routing and capacity In our implementation, the connections cannot form loops Graphs are activated and deactivated by a management function Similarities to ATM circuits, flow labels, and Cisco tag switching (MPLS) 27

Superframe A superframe is a collection of slots that repeat periodically in cycles Binds/implements links from a graph with specific time/channel offset Superframe Unallocated Slot Allocated Slot 28

Link = (Time Slot, Channel Offset) D One Slot Time Chan. offset A B A C D A B B A B C E F B E B F All of B’s transmit links are dedicated and won’t collide – B has twice as much bandwidth to A as to C – B can broadcast to E and F, or use that link of a unicast to one or the other D and C share a link for transmitting to A. A backoff algorithm is needed in case of collisions. 29

Low latency – high reliability Dedicated bandwidth for first k transmissions Shared bandwidth for next j All transmissions are on different channels DR ~= 1 -PERk+aj e. g. PER=0. 1, 4 inputs 100 ms max latency K=2, J=4 ~99. 9999% 30

Low latency – multiple hops Color indicates TX slot for mote of that color 5 hops = 5 slots = 50 ms Can repeat immediately, or every k slots, according to superframe length. Peak throughput with n motes in a line: 1 payload / (n*10 ms) = 100/n payloads/second 31

High throughput – multiple hops Use multiple channels Odd slots Only need 2 time slots Throughput is 50 packets/second – independent of n Even slots – Limited by worst PER (not combination) Odd slots Even slots Odd slots 32

Source routes, broadcast, multicast Source routing 17 – Contains graph ID – Can transition to flood Broadcast 47 – Flood along graph – MAC & NET broadcast ID Multicast 1 41 53 80 – MAC broadcast ID – NET multicast ID 43 Multicast 2 81 82 – Broadcast on graph covering multicast group 33

Performance Limits Data collection – 100 pkt/s per gateway channel – 16*100 pkt/s with no spatial reuse of frequency Throughput – ~80 kbps secure, reliable end-to-end payload bits per second per gateway – 16 * 80 k = 1. 28 Mbps combined payload throughput w/ no spatial reuse of frequency Latency – 10 ms / PDR per hop – Statistical, but well modeled Scale – > 1, 000 nodes per gateway channel 34

50 motes, 7 hops 3 floors, 150, 000 sf >100, 000 packets/day

36

37

38

RF Path Loss – R 2 ? R 4 ? 39

40

Path Stability by 802. 15. 4 Channel 802. 11 bg Channels 802. 15. 4 Channels 2. 480 GHz 2. 405 GHz Each dot is the 15 minute average 41

42

43

44

Oil Refinery – Double Coker Unit Scope limited to Coker facility and support units spanning over 1200 ft No repeaters were needed to ensure connectivity Electrical/Mechanical contractor installed per wired practices >5 year life on C-cell GW 400 m 45

Barriers to Adoption >99. 9% Wireless HART, SP 100 “It just worked” 5 -10 years Complete networks On. World, 2005 46

Excerpts from Presentations at the Emerson Process Users Conference September 2007 47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

Conclusion Wireless Sensor Networks for Industrial Automation about to take off Based on Berkeley/BSAC-dervived TSMP Other markets and standards will follow 62

Multiple Applications in Multiple Markets Industrial automation Building automation Defense & security Wireless-enabled applications 63

Industrial Monitoring Emerson’s Smart. Wireless products based on Dust’s products 12 companies representing 90% of the market for industrial field devices demonstrated interoperable prototype products based on Dust products "Wireless promises to enable us to put more monitoring in the plant at one-tenth the cost of wired technology” John Berra, President Save up to 90% on installation The cost of wire, additional hardware and labor drives up the cost of any project, large or small. Wireless solutions enable costeffective implementation of new measurement points. Emerson Wireless Engineering Over Wired Installation Other 90% Savings 0% 10% Materials Labor 100% 64

Oil and Gas “…this wireless technology enabled us to do things we simply could not do before, either because of cost or physical wiring obstacles. Through the trials, we found that Emerson's wireless approach is flexible, easy to use, reliable, and makes a step-change reduction in installed costs. " Dave Lafferty BP “Wireless truly is faster and cheaper. “ “It just worked!” Brandon Robinson En. Cana Savings of a 5 -node installation: 700’ conduit 3000’ wire 2 guys, 2 full days of labor no trenching or surveying for buried cable 65

Predictive Maintenance "Unscheduled downtime is the largest single factor eroding plant performance. Over $20 Billion, or almost 5 percent of total production, is lost each year in North America alone due to unscheduled downtime. " ARC, 2002 “Electric motors consume approximately 60% of all electricity generated in the United States. “ US Do. E, December 2002 Ubiquitous monitoring of motors, pumps, and bearings: Vibration Temperature Acoustic 66

Parking Monitoring Real-time monitoring of parking for: Increased enforcement Dynamic pricing Real-time vacancy location services Wireless sensor node 67

Perimeter Security Monitoring for perimeter violations: Ground vibration (footfalls or vehicles) Metal (vehicles) Sound Motion 68