Wireless Sensor Networks WSN 1 Comparing to

Wireless Sensor Networks (WSN) 1

Comparing to MANET • Similarities – No infrastructure – Multi-hop communication • Differences – – – Nodes are more resource-constrained and more prone to failure More nodes (up to hundreds or thousands) in a network Random deployment Unattended Longer life time Trust relationships between sensor nodes (typically belong to the same organization) – Application-specific – … 2

Usage of Sensor Networks • Environmental observation – Water/air pollution detection – Forest fire detection – Animal habitat monitoring • Military monitoring – Battlefield surveillance – Vehicular traffic monitoring – Tracking the position of the enemy • Building monitoring – Monitoring climate changes/vibration • Healthcare – Being implanted in the human body to monitor medical problems 3

Some Real Examples • Great Duck Island – A prototype sensor network is deployed to monitor the nesting grounds of elusive seabirds – Biologist get information they need with minimal human disturbance • Vineyard – Embedded sensors are deployed to monitor temperature in a vineyard in Oregon’s Willamette Valley. • Golden Gate Bridge – 200 motes organized in an ad hoc sensor network are used for tracking stress on the bridge • Proactive Health Research Project (Intel) – Help seniors age with dignity and independence, by developing sensor network-based in-home technology prototypes. • Preventive maintenance on an oil tanker in the North Sea (Intel and BP) 4

Sensor Nodes • MICA 2(Motes): a popular research platform at the moment (J. Hill, et al. , “The platforms enabling wireless sensor networks, ” Comm. of the ACM, June 2004/Vol 47. No. 6, pages 41 -46) 5

Sensor Nodes: MICA 2 • The core is a small, low-cost, low-power computer – Atmel Atmega 128 L processor (4 MHZ), 128 KB on-board flash memory – As powerful as 8088 CPU (in original IBM PC) – Power consumption • 8 milliamps (running), 15 micro-amps (sleep) • One or more sensors can be mounted • Connect to the outside world with radio – Transmission range: 10 -200 feet, Rate: 76 bps – Power consumption • 25 milliamps (Trans), 10 milliamps (Recv), <1 milliamps (off) • Power supply: 2 AA batteries – 2, 000 milliamps-hours (http: //computer. howstuffworks. com/mote. htm) 6

Sensor Nodes • Four classes (J. Hill, et al. , “The platforms enabling wireless sensor networks, ” Comm. of the ACM, June 2004/Vol 47. No. 6, pages 41 -46) 7

Physical layer and MAC sublayer • Related Standard – IEEE 802. 15. 4 Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (LR-WPANs) • Physical layer – 2. 4 -2. 4835 GHZ (worldwide) – 902 -928 MHz (North America) or 868 -868. 6 MHz (Europe) • MAC Sublayer – Contention Access • CSMA-CA – Contention Free 8

Other MAC Protocols (not complete) Protocol Summery PAMAS Power Aware Multi Access protocol with Signaling for Ad Hoc Networks: uses two separate channels, one for control packets and one for data packets. It puts the nodes to sleep when the neighbor is transmitting. S-MAC Periodic Listen and Sleep, Collision and Overhearing Avoidance, Message Passing, Implemented over Berkeley Mote. TDMAAssumes a gateway per cluster where the gateway assigns based MAC the time slots for each node. protocol Adaptive Rate Control Random Delay Good for per node fairness. Uses adaptive rate control to adapt the originating data and route-thru traffic. Rate control mechanism uses linear increase and multiplicative decrease where each one. Their CSMA scheme uses sensor sampling 9 phase shift to avoid capturing effect.

Deployment • Random deployment – Large number of nodes; target area may be remote and/or hostile manual deployment is impossible in many cases • Problems caused by random deployment – Localization • Sensor nodes must discover their locations after deployment – Coverage • For sensing quality, a certain level of sensing coverage should be achieved. – Security • It is hard to store various encryption keys on nodes, since the neighborhood cannot be know a priori. 10

Localization • Basic idea of existing schemes – Initial configuration • A small number of beacon nodes – Know their locations by using GPS or being set manually • A large number of nodes (non-beacon nodes) that do not know their locations – Localization process • Beacon nodes send beacon signals to a set of non-beacon nodes • A non-beacon node obtains – Locations of the beacon nodes – Some features related to the distance to these beacon nodes » Received signal strength indicator, Time to arrival, etc. • The non-beacon node estimate its own location based on the obtained information 11

Sensing Coverage • Objectives – High sensing coverage (full coverage is ideal) – Energy efficient and low cost • Samples of existing work – Sleep scheduling (Ye et al. 03’, Gui et al. 04’) • Over deploying sensor nodes and make spare nodes sleep • Waking up sleeping nodes following certain sleep planning methods – Employing mobile sensor nodes (Wang et al. 03’&05’) • Deploy a large fraction of stationary nodes and a small fraction of mobile nodes • Stationary nodes detect sensing holes and notifies mobile nodes • Mobile nodes move to heal the holes 12

Network Architectures: Flat Sink (data collector) • A large number of sensor nodes form a peer-to-peer ad hoc network • They forward message for each other 13

Network Architecture: Hierarchical Sink Cluster head • The network is divided into clusters, each cluster has a head (logically or physically) • Ordinary node cluster head; cluster heads form a backbone 14

Network Architecture: Hierarchical (J. Hill, et al. , “The platforms enabling wireless sensor networks, ” Comm. of the ACM, June 2004/Vol 47. No. 6, pages 41 -46) 15

Communication Patterns • A sensor network is a content-based (or datacentric) network – In WSN, networking take place directly on contents (or data) • In Internet/MANET, networking protocols use identifiers of nodes. – Contents can collected, processed and stored in the network • Desirable interaction paradigm in WSN: Publish/Subscribe – Entities can publish data under certain names – Entities can subscribe to updates of such named data (H. Karl, “Ad hoc and sensor networks Chapter 12: Data-centric and content-based networking”) 16

Naming • Content based naming – Tasks are named by a list of attribute – value pairs – Task description specifies an interest for data matching the attributes • Animal tracking: Request Interest ( Task ) Description Type = four-legged animal Interval = 20 ms Duration = 1 minute Location = [-100, -100; 200, 400] Reply Node data Type =four-legged animal Instance = elephant Location = [125, 220] Confidence = 0. 85 Time = 02: 10: 35 17

Communication Patterns • External storage-based pattern – fixed subscriber • Sink-initiated pattern – subscriber initiated • Source-initiated pattern – publisher initiated • Rendezvous-based pattern – Intermediate entities help to match publishers and subscribers 18

External Storage-based Pattern source sink 19

Sink-initiated Pattern: Directed Diffusion • The sink periodically broadcasts interest messages to each of its neighbors • Every node maintains an interest cache – Each item corresponds to a distinct interest – No information about the sink – Interest aggregation : identical type, completely overlap rectangle attributes • Each entry in the cache has several fields – Timestamp: last received matching interest – Several gradients: data rate, duration, direction (Intanagonwiwat et al. , “Directed Diffusion for wireless sensor networking, ”, IEEE/ACM Transactions on Networking (TON), Feb, 2003) 20

Directed Diffusion • Set up gradient: Constrained or Directional flooding based on location. Gradient Source Interest Sink (Intanagonwiwat et al. , “Directed Diffusion for wireless sensor networking, ”, IEEE/ACM Transactions on Networking (TON), Feb, 2003) 21

Directed Diffusion Gradient Source Data Sink (Intanagonwiwat et al. , “Directed Diffusion for wireless sensor networking, ”, IEEE/ACM Transactions on Networking (TON), Feb, 2003) 22

Source-initiated Patterns Dissemination Node Data Announcement Source Data Sink Query Immediate Dissemination Node (Ye et al. , “TTDD: A Two-tier Data Dissemination Model for Large-Scale Wireless Sensor Networks, ” Mobicom’ 02) 23

Rendezvous-based pattern • Data with the same name are stored at the same place (Ratnasamy et al. , “Data-centric Storage in Sensor Networks with GHT, ” Mobile Networks and Applications, 2004. ) 24

Rendezvous-based pattern • Fault tolerance consideration • If a storing fails, it is replaced by another node closest to itself • To protect the stored data, data can be replicated in multiple nodes (Ratnasamy et al. , “Data-centric Storage in Sensor Networks with GHT, ” Mobile Networks and Applications, 2004. ) 25

Routing • Address-centric (AC) protocol – Each source independently sends data along the shortest path to sink based on the route that the queries took (“end-to-end routing”) • Data-centric (DC) protocol – The sources send data to the sink, but routing nodes en-route look at the content of the data and perform some form of aggregation/consolidation function on the data originating at multiple sources. (Krishnamachari et al. , “Modeling data-centric routing in wireless sensor networks”, 2002. ) 26

Data Aggregation • Types – Removing redundancy – In-network processing After aggregation, the redundancy is removed Sensing range of A Redundancy 27 Sensing range of B

Data Aggregation: in-network processing • Aggregation is implemented by three functions – Merging function f, initializer i and evaluator e. • General form = f (, ) – and are multi valued partial state records – is partial state record resulting from application of f to and f(<3, 1>, <5, 1>) = <8, 2> • E. g. f is the merging function of AVERAGE f(<8, 2>, <7, 1>) = <15, 3> e(<15, 3>) = 15/3 = 5 f(, )= i(x) = e() = S/C where S and C are Sum and Count. <3, 1> i(3)=<3, 1> i(7)=<7, 1> <5, 1> i(5)=<5, 1> 28