Plan Introduction Sensor networks [Estrin, Mobicom 1999] Information gathering and processing Data centric: data is requested based on certain attributes Application specific 1. I ntroduction (10 min.) Energy constraint Real-time I ssues in Wireless Sensor Data aggregation (also data fusion) 2. MAC layer (10 min.) Networks Military applications: (4C’s) Command, control, communications, computing 3. Network layer (15 min.) Intelligence, surveillance, reconnaissance Targeting systems David SI MPLOT-RYL Health care 4. Other issues (8 min.) Monitor patients I RCI CA/ LI FL, CNRS UMR 8022, Université de Lille 1 Assist disabled patients I NRI A Futurs, France Commercial applications 5. Conclusion (2 min.) Managing inventory Monitoring product quality http://www.lifl.fr/~simplot Monitoring disaster areas simplot@lifl.fr ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks Sensor Nets for Sensor Nets for Sensor Nets for Search and Rescue Search and Rescue Search and Rescue Active Sensor Inactive Sensor ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 1
Application Event-driven model Introduction Design factors [Akyildiz et al. I n the UC Berkeley Botanical Garden, 50 “micromotes” 2002] sensors are dangled like earrings from the branches of 3 Fault Tolerance (sustain functionalities) redwood trees to monitor their growth. Scalability (hundreds or thousands) Production Cost 25 Motes on (now $10, near future $1) Damaged sidewall Hardware Constraints Network Topology (pre-, post-, and re-deployment) Transmission Media (RF, Infrared, and Optical) Temperature Power Consumption Seismic wave (with < 0.5 Ah, 1.2 V) Passive sensors RFID Tags that can provide context- aware information Sensor nodes ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks Ad hoc networks vs On-demand model RFID Tags sensor networks Sensor networks are simplified versions of ad hoc networks Smart labels Mono-application Radio Frequency Identification Tag Communication By opposition to bar code which use optical � 1-to-all restricted to sink-to-sensors principles � 1-to-1 restricted to sensor-to-sink A stronly limited component: 500 times smaller than a classical Sensor networks need more than ad hoc networks microprocessor Communication � All-to-1 for sensors-to-sink Sensor networks need less than ad hoc networks It does not matter if a node dies because of energy consumption Intel Pentium 4 Global application has priority on life of nodes Chip with a size of some mm² Sensor networks are not ad hoc networks RFID Tag It uses same tools and theory but it is different ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 2
Principle EAS Application Applications Electronic Article Surveillance Batch identification Typically, RFID Tags car passive components: they have no Once powered, the tag emits It is the capability to collect information from a set of tags battery! The reader listen channel and activate alarm as early as transmission is In opposition to optical identification optique Tag are powered by electromagnetic field generated by reader detected Communication from reader device to vicinity tags: amplitude shift keying (ASK) During checkout, the tag is burner out Communication from tags to reader device: impedance shift keying (ISK) Problem: aliment the tag whatever the tag orientation Marathon Automatic clocking in Automatic luggage sorting Automatic inventory 50 items in less than one second courtesy Intersoft ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks Networking the More POPS, smaller POPS… physical world 100 µ m Savant RF Tag Control System Courtesy, Auto-ID Center The MIT Auto-ID Center Vision of “the Courtesy, Alien Technology internet of things” Networked Tag Readers ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 3
Real-Time Constraints Plan 2. MAC Layer I n some applications What is MAC layer? Déjà vu ! Healthcare, fire surveillance, etc. Generally assimilated to link layer Its role is to schedule packets and to deal with collisions delay delivery of alarm is a natural constraint 1. I ntroduction Network layer 2. MAC layer OSI � Outgoing queue Incoming queue 3. Network layer Network Layer 3 FIFO � � 4. Other issues Data Link Layer 2 � � MAC Protocol SendPacket Sense Physical 5. Conclusion Layer 1 WaitACK ReceivePacket Physical layer ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks Family of Non- One-Hop Delay Management Inter-Node Differentiation Deterministic Protocols Basic tricks: Two distinct families: Example of random access protocol: 802.11 Network layer labels packets with deadline and packet scheduler uses Random access protocols / non-deterministic protocols EDF (multi-hop estimation? Cross-layer with higher layers) � 802.11 WiFi Introduction of several priority classes with different outgoing queues Contention free protocols / deterministic protocols History It introduces internal packet differentiation � Time Division Multiple Access (TDMA) � E.g. alarm packet has priority on control packet PNAV � 802.11 � � � � � S-MAC ALOHA CSMA MACA AOB Collision MACAW 1970 1980 1990 2000 2010 � Internal differentiation No inter-node differentiation ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 4
Collision-Based MAC Interframe Slots Backoff Procedure Protocols Start Priorities are defined through different inter frame spacing ALOHA: Emit message SI FS (Short I nter Frame Spacing) Highest priority, for ACK,CTS,polling response Random wait Déjà vu ! packet radio networks PI FS (PCF I FS) ACK ? send when ready medium priority, for time-bounded service using PCF timeout or NACK 18-35% channel utilization DI FS (Distributed Coordination Function I FS) yes Lowest priority, for best-effort data End DIFS DIFS CSMA (Carrier Sense Multiple 26 % PIFS Access): 37 % SIFS medium busy contention next frame “listen before talk” t Déjà vu ! direct access if 50-80% channel utilization 37 % medium is free ≥ DIFS Déjà vu ! ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks Backoff Procedure Family of Deterministic Introduce Priority (continued) Protocols For each priority class I nitial contention window size Classical TDMA (like in GSM) DIFS high priority = short DIFS CW = CWmin (32 slots) CWmin high priority = small CWmin Déjà vu ! Déjà vu ! CWmax high priorty = small CWmax Backoff is randomly chosen Slot0 Slot1 Slot3 Slot4 Slot5 Slot6 Slot7 Slot0 Slot1 m high priority = small m backoff = rand(0..CW) Node A DIFS DIFS Node B PIFS Each time that a collision occurs SIFS � i.e. Backoff=0 for at least two neighbor nodes medium busy contention next frame Node C CW = min( m × CW, CWmax ) (m=2, CWmax=1024) Node D Reminder Node E Initial contention window size CW = CWmin Each time that a packet is sent successfully backoff = rand(0..CW) CW = CWmin Each time that a collision occurs CW = min(CWmax, m×CW) Variation (linear decrease): CW = max( CWmin , CW- CWmin ) Each time that a packet is sent successfully CW = CWmin ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 5
TDMA: Pros and Cons Unit Disk Graph TDMA: Pros and Cons Pro: Pro: Collision free Collision free Estimation of delay is easier (see later) Estimation of delay is easier (see later) Cons: Cons: Optimal Slot number depends on maximal degree of the graph Optimal Slot number depends on maximal degree of the graph � Time may be wasted in sparse zones � Time may be wasted in sparse zones Initialization may be long � Graph coloring of 2-hop graph � Well-known disadvantage of SMAC Time synchronization is required ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks One-Hop Delay Estimation Plan 3. Network Layer Measurement We limit the presentation to unicast routing With time synchronization � Simple dedicated hand check 1. I ntroduction Three different approaches � Or average in exchanged messages Table driven routing protocols Without time synchronization On demand routing protocols 2. MAC layer � Hand check and delay is assumed to be symmetrical Position-based routing protocols False assumption in general 3. Network layer Problem: delay is known to vary with message length Estimation 4. Other issues Modeling of MAC layer (e.g. M/G/1) Average delay Delay probability distribution 5. Conclusion Can take message length into account, number of interfering nodes, etc. Problem: ignore software execution (WCET?) ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks ETR05 SIMPLOT-RYL – Sensor networks 6
Recommend
More recommend
