Mobile Networks Module G – Wireless Sensor Networks

1. Mobile NetworksModule G – Wireless Sensor Networks J.-P. Hubaux and J. Panchard

2. Outline • What technology for what applications? • WSNs characteristics and design issues, with special focus on: • Power management • Reliable data collection • Hybrid architectures • Time Synchronization • Conclusion

3. The Internet revolution was about people wanting to talk to each other. Now we realize that things have much to tell as well.

4. Motivation Earthquake detection • The occurrence of an earthquake can be detected automatically by accelerometers. • Earthquake speed: around 5-10km/s • If the epicenter of an earthquake is in an unpopulated area 200km from a city center, an instantaneous detection system can give a warning up to 30 seconds before the shockwave hits the city. • If a proper municipal actuation network is in place: • Sirens go off • Traffic lights go to red • Elevators open at the nearest floor • Pipeline valves are shut • Even with a warning of a few seconds,the effects of the earthquake can bemitigated. • Similar concept can be applied to • Forest fire • Landslides • Etc. Ref: RagavSZ06

5. Motivation Cold Chain Management • Supermarket chains need to track the storage temperature of perishable goods in their warehouses and stores. • Tens if not hundreds of fridges should be monitored in real-time • Whenever the temperature of a monitored item goes above a threshold • An alarm is raised and an attendant is warned (pager, SMS) • The refrigeration system is turned on • History of data is kept in the system forlegal purpose • Similar concept can be applied topressure and temperature monitoring in • Production chains • Containers • Pipelines Ref: http://www.ip01.com

6. Motivation Home automation • Temperature management • Monitor heating and cooling of a building in an integrated way • Temperature in different rooms is monitored centrally • A power consumption profile is to be drawn in order to save energy in the future • Lighting management: • Detect human presence in a room to automatically switch lights on and off • Responds to manual activation/deactivation of switches • Tracks movement to anticipate the activation of light-switches on the path of a person • Similar concept can be applied to • Intrusion detection

7. Motivation Precision Agriculture management • Farming decisions depend on environmental data (typically photo-synthesis): • Solar radiation • Temperature • Humidity • Soil moisture • These data evolve continu-ously over time and space • A farmer’s means of action to influence crop yield : • Irrigation • Fertilization • Pest treatment • To be optimal, these actions should be highly localized (homogenous parcels can be as small as one hectare or less) • Environmental impact is also to be taken into account • Salinization of soils • Groundwater depletion • Well contamination

8. Motivation WaterSense • Goal: Help define and implement farming strategies for farmers in a situation of water scarcity. • Crop assessment • Water conservation measures • Time of farming operations • Real-time monitoring of the field conditions • Desired Outcome: farming decision support system based on environmental data Ref: Panchard06

9. Motivation Ubiquitous computing economic forecast • Industrial Monitoring (35% – 45%) • Monitor and control production chain • Storage management • Monitor and control distribution • Building Monitoring and Control (20 – 30%) • Alarms (fire, intrusion etc.) • Access control • Home Automation (15 – 25%) • Energy management (light, heating, AC etc.) • Remote control of appliances • Automated Meter Reading (10-20%) • Water meter, electricity meter, etc. • Environmental Monitoring (5%) • Agriculture • Wildlife monitoring • Other areas: • Performance monitoring in sports • Patient monitoring in health and medicine • Sensor going wireless in vehicular networks

10. WSN Characteristics and design issues • Characteristics • Distributed data collection • Many-to-one (rarely peer-to-peer) • Limited mobility • Data collection (time and space resolution) • Event detection • Minimal intrusiveness • Design issues • Low-cost (hardware and communication) • Extended life-time • Reliable communication • Efficient integrated data processing • Hybrid network infrastructure • Security

11. Characteristics Wireless Sensor Network architecture • Numerous sensor devices • Modest wireless communication, processing, memory capabilities • Form Ad Hoc Network (self-organized) • Report the measured data to the user

12. Characteristics Sensor interface sensor RF transceiver Signal processing Actuator interface actuator Data processingand storage Sensor Node architecture • A sensor node can be an information source, a sink and a router • Autonomous low-power • Combine sensing, signal conditioning, signal processing, control and communication capabilities (courtesy of Swiss Center for Electronics and Microelectronics, Neuchâtel)

13. 868 MHz multi-channel transceiver 8 MHz μ-Controller 10KB RAM 48 kB Program space 512 External Flash 115 kbps data rate 3 V supply voltage Current consumption Transmit 33 mA Receive 14 mA Sleep < μA -121 dBm sensitivity Radio range 200m (outdoor) 39 MHz quartz reference Characteristics Example of a low power transceiver:TinynodeTM Tinynodes are a registered trademark of Shockfish, Ecublens

14. Design Issue: Low-cost • Hardware • Low-cost radio • Low-cost internal clock • Limited storage and processing capabilities • Not tamper-proof • Communication • Cannot rely on existing pay-per-use cellular infrastructure

15. Design Issue: Power Management • Energy-efficient routing • Load-balancing • Mobility • In-network aggregation • Medium-access control

16. Design Issue: Power Management Simple radio model d ETx(k, d) ERx(k) Transmitelectronics Receiveelectronics Tx Amplifier k bit packet k bit packet Eamp * k * da Eelec * k Eelec * k ERx(k) = Eelec * k ETx(k, d) = Eelec * k + Eamp * k * da Typical values:a = 2…6 Eelec = 50 nJ/bit Eamp = 100 pJ/bit/ma

17. Design Issue: Power Management Load-balancing • Assumption: in a multi-hop many-to-one sensor network, the data collection follows a spanning tree. • In most cases, power consumption per node due to transmission/reception grows exponentially from the leaves to the root of the tree • Consequence: the batteries of the nodes close to the sink deplete faster. Since they relay all the network’s traffic, they pull the network lifetime down.

18. Design Issue: Power Management Load-balancing Line topology Tree topology o 1 2 1 3 … N BS … N

19. Design Issue: Power Management Load balancing • Power consumption increases at least linearly when nodes are closer to the sink • Typical case is much worse Ref:LuoH05

20. Design Issue: Power Management Use mobility for load-balancing • Move the base station to distribute the role of “hot spots” (i.e., the nodes around the base station) over time • The data collection continues through multi-hop routing wherever the base station is, so the solution does not sacrifice latency Ref:LuoH05

21. Design Issue: Power Management In-network data aggregation • To mitigate cost of forwarding • Compute relevant statistics along the way: mean, max, min, median etc. • Forwarding nodes aggregate the data they receive with their own and send one message instead of relaying an exponentially growing number of messages • Issues • Location-based information (which nodes sent what) is lost • Distributed computation of statistics • mean: node needs to know both the mean values and the sizes of samples to aggregate correctly • median: only an approximated computation is possible • Especially useful in a query-based data collection system • Queries regard a known subset of nodes • Aggregation function can be specified

22. Design Issue: Power Management Medium-Access Control • MAC attributes: • Collision avoidance • Energy efficiency • Scalability and adaptation • Nodes transmit very intermittently, but once a transmission is taking place, we must ensure that the intended receiver gets it. • Current-consumption in receive state or in radio-on idle state are comparable • Idle state (idle listening) is a dominant factor in power consumption Trying to put nodes to sleep most of the time, and wake them up only to receive a packet

23. Design Issue: Power Management Synchronous MACs • TDMA (cellular networks) • Shortcomings • Necessity to organize nodes in clusters and cluster hierarchies • High control traffic cost • Possible solution • Each node maintains two schedules • Its parent schedule • The schedule it sets for its children • Beacons are used to compensate for clock drifts … … 1 2 3 N 1 2 3 N Frame 2 Frame 1 Ref:BurriRW07

24. Design Issue: Power Management Asynchronous: B-MAC • Asynchronous • Low Power listening Awake Preamble Data Receiver On Off Sender |off period| ≈ |preambule| Ref: PolastreHC04

25. Design Issue: Power Management Shortcomings • Transmitting a packet is very expensive • Overhearing is expensive • Relaying packets is expensive (multihop) Simple Improvement: • Aggregating packets before sending them • In low duty cycle data collection network, gain may be substantial • Price to pay : real-time

26. Design Issue: Reliable data collection • Many-to-one communication paradigm • Multi-hop communication • Nodes select one parent to send their data packets (tree topology)

27. Design Issue: Reliable data collection MintRoute: A data collection tree routing protocol • A distance-vector routing protocol • One routing path per node • A shortest path: Minimum number of transmissions • The base station send periodic beacons that are broadcast by each node after incrementing a hop count • Nodes select beacons with lowest hop count from the ones it received, and adds its sender among a list of potential parents • Neighboring nodes exchange periodic beacons for link quality evaluation (gaps within the sequence # of packets → packet losses) • Nodes select their parent based on hop count, link quality and load. • Volatile routing topology → load balancing • Cycle avoidance : Link quality must not be allowed to vary to rapidly Ref:WooTC03

28. Design Issue: Reliable data collection MintRoute : root beaconing 2 (1) 7 3 (8) 2 (1,3) 8 13 2 (2) 6 1 1 1 3 1 2 Sink 9 0 2 (4) 1 3 (11) 4 12 1 14 5 3 (9) 10 11 2 (4,5) 2 (5)

29. Design Issue: Reliable data collection MintRoute : link estimation 2 (1) 7 3 (8) 2 (1,3) 8 13 2 (2) 6 1 1 1 3 1 2 9 0 2 (4) 1 3 (11) 4 12 1 14 5 3 (9) 10 11 2 (4,5) 2 (5)

30. Design Issue: Data Processing • Challenge: develop a middle-ware that makes it convenient to plug any wireless sensor network to a global querying space. • Cope with • Data heterogeneity • Geographic distribution • Platform diversity • Etc… • Web-based interface to • Add/remove instances of sensor networks • Display data • Export data • Send queries • Example: http://gsn.sourceforge.net/

31. Design Issue:Hybrid network architecture • Deployment can be in remote areas • The sink needs to communicate the data to the user • Pervasive computing means pervasiveness of access to data • Multi-tier architectures are necessary

32. Design Issue: Hybrid network architecture 2-tier architecture with 802.11 bridge

33. Design Issue: Hybrid network architecture Delay Tolerant network with data mules • Clusters are not directly connected to the server • Cluster heads store data from the cluster nodes • “Data mules” collect the data periodically • When a cluster-head detects a mule, it uploads to it the data it had in store

34. Need for time synchronization common reference points between neighbors at sensing to obtain a global time-stamp synchronized sensing and actuation time arithmetic in local and global time Classification of algorithms Global time source: internal vs. external Lifetime: on-demand vs. continuous Scope: all nodes vs. subsets vs. pairs Time transformation: local clock adjustment vs. local and global time pair vs. timescale transformation Clock drift: offset vs. skew and offset compensation Local clock: CPU (high resolution, no power management, not stable) vs. external crystal Method: sender–receiver vs. receiver–receiver Time stamping: MAC layer vs. user space Approaches Network Time Protocol (NTP) Reference Broadcast Synchronization (RBS) Timing-sync Protocol for Sensor Networks (NTSP) Flooding Time Synchronization Protocol (FTSP). Metrics It is NOT only end-to-end accuracy Network load (in messages per seconds per nodes) Start-up time (as the function of the diameter) Fault tolerance nodes entering and leaving incorrect and unstable local clocks topology changes Scalability Time synchronization Ref:Maroti04

35. Time synchronization Radio message propagation • byte alignment time: delay incurred because of different byte alignment of the sender and receiver • access time: delay incurred waiting for access to the transmit channel • interrupt handling time: The delay between the radio chip raising and the microcontroller responding to an interrupt • encoding time: the time it takes the radio chip to encode and transform a part of the message to electromagnetic waves • propagation time: time it takes for the electromagnetic wave to travel from sender to receiver • decoding time: the time it takes on the receiver side to transform and decode to binary representation

36. Time synchronization FTSP time stamping • Time synchronization primitive: establishing time reference points between a sender and receiver(s) using a single radio message • Sender obtains timestamp when the message was actually sent in its own local time • The message can contain the local time of the sender at the time of transmission (or the elapsed time since an event) • Receiver obtains timestamp when the message was received in its own local time • Algorithm • Multiple time stamps are made both on the sender and receiver sides at byte boundaries • The time stamps are normalized, and statistically processed and a single time stamp is made both on the sender and receiver sides • The final time stamp on the sender side can be embedded in the message • Uses • time synchronization protocols • acoustic ranging • shooter localization (implicit time synchronization while routing) R1 S R3 R2 transmission / reception event S elapsed time since event R1 R2 R3 time stamping error

37. Time synchronization Time stamping on MICA2 platform header data sender … sync time stamp 0 1 2 3 4 5 hw and sw delay (~1386 μs) bit-offset (~0-365 μs) 0 1 2 3 4 5 byte (~417 μs) hw interrupt sw interrupt receiver handling delay (95% 0-1 μs) (5% 1-20 μs) min min average MICA2: 1.4 μs average error, 4.2 μs maximum error MICA2DOT: 4 μs average, 12 μs maximum error Limiting factor: the stability of the CPU clock

38. Time synchronization FTSP time synchronization • Local and global time • Each node maintains both a local and a global time. Past and future time instances are translated between the two formats • Both the clock offset and skew between the local and global clocks are estimated using linear regression • Optimized to increase the numerical precision of calculations by normalizing the clock skew to zero (working with skew minus one) • Handles local and global clock overflows (uses 32-bit integers) • Hierarchy • Global time is synchronized to the local time of an elected leader • Utilizes asynchronous diffusion: each node sends one synchronization msg per 30 seconds, constant network load • Sequence number, incremented only by the elected leader • Continuous operation, startup time is network diameter times 60 seconds • Robustness • If leader fails, new leader is elected automatically. The new leader keeps the offset and skew of the old global time • Fault tolerant: nodes can enter and leave the network, links can fail, nodes can be mobile, topology can change • Platform independent: uses the time stamping module 15 16 17 15 16 16 17 leader 17 16 15 17 16 15 16 14 16 15 14 15 sequence numbers

39. Time synchronization 1st leader 2nd leader FTSP experimental evaluation topology: avg. error: 1.6 μs max. error: 6.1 μs per hop 50% turned off 1st leader turned off random nodes turned off/on all turned back on all turned on

40. Time synchronization FTSP Contributions • Better understanding of the uncertainties of radio message delivery and new techniques to reduce their effect • FTSP time stamping (single-hop time synchronization) • Uses a single radio message to synchronize sender and receiver(s) • The most basic time synchronization primitive • 1.4 μs average, 4.2 μs maximum error on the MICA2 • Implemented on MICA2, MICA2DOT and MICAZ • FTSP multi-hop time synchronization • Synchronizes all nodes to an elected leader • Constant network load: 1 message per 30 seconds per node • Continuous operation, startup time is network diameter times 60 seconds • Fault tolerant: nodes can enter and leave the network, links can fail, nodes can be mobile, topology can change • Platform independent: uses the time stamping module

41. Reference Broadcast Synchronization (RBS) Uncompensated delays: propagation, decoding, byte alignment, interrupt handling and receive times Network overhead: 1.5 msgs per synchronization period Hierarchy: clustered with timescale transformation Timing-sync Protocol for Sensor Networks (TPSN) Uncompensated delays: encoding, decoding, interrupt handling times Network overhead: 2 msgs per synchronization period Hierarchy: spanning tree Time synchronization Comparison to RBS and TPSN • Flooding Time Synchronization Protocol (FTSP) • Uncompensated delays: propagation time • Network overhead: 1 msg per synchronization period • Hierarchy: flooding • End-to-end accuracy: average 1.6 μs per hop, max 6.1 μs per hop on MICA2 • Robustness: nodes can enter and leave the network, links can fail, nodes can move • Startup: 2x diameter many synchronization periods • FTSP time stamping • singe radio message • 0 or 4 bytes overhead per msg • 1.4 μs average, 4.2 μs maximum error on MICA2

42. Time synchronization FTSP Conclusion • Time stamping: new time synchronization primitive • time synchronization protocols • synchronized sensing / actuation • power management • CPU clock cycle precision • the effects of discrete time • separate local and global times • how to sense and actuate with this precision • Robustness and scalability • startup and convergence time • scaling to 10000 nodes • power management • One size does not fit all • need a spectrum of time synchronization algorithms • application specific and integrated solutions

43. General conclusion on sensor networks • WSNs are an emerging technology which is promised to a dramatic growth in the years to come • This new paradigm of communication represents specific design constraints • They must be extremely low-cost (both to purchase and to operate) • They must be extremely energy efficient since their lifetime is potentially years • Hardware design • Routing and topology mechanisms • Specialized Medium Access Control mechanisms • Despite their low-cost and power management features, they must implement reliable communication protocols • They must integrate versatile middle-ware and provide data processing • They will often rely on a hybrid network infrastructure

44. References [BurriRW07 ] N. Burri, P. von Rickenbach and R. WattenhoferDozer: Ultra-Low Power Data Gathering in Sensor NetworksIPSN, 2007 [LuoH05] Jun Luo and Jean-Pierre HubauxJoint Mobility and Routing for Lifetime Elongation in Wireless Sensor NetworksINFOCOM'05 [Maroti04] M. Maroti, B. Kusy, G. Simon, A. LedecziThe Flooding Time Synchronization Protocol ACM SenSys, 2004 [Panchard06] J. Panchard, S. Rao, T.V. Prabhakar, H.S. Jamadagni and J.-P. Hubaux, Improved Water Management for Resource-Poor Farmers via Sensor Networks, ICTD, 2006 [PolastreHC04]J. Polastre, J. Hill and D. Culler Versatile Low Power Media Access for Wireless Sensor Networks SenSys, 2004 [RaghavSZ06] C.S. Raghavendra, K.M. Sivalinguam and T. Znati EditorsWireless Sensor NetworksSpringer, 2006 [WieselNE01] J.E. Wieselthier, G.D. Nguyen, and A. EphremidesAlgorithms for energy-efficient multicasting in static da hoc wireless networksACM MONET, June 2001 [WooTC03] Alec Woo, Terence Tong, and David CullerTaming the Underlying Challenges of Reliable Multhop Routing in Sensor NetworksSenSys 2003