1 Application Event-driven model Introduction Design factors - - PowerPoint PPT Presentation

1

SIMPLOT-RYL – Sensor networks

ETR05

Real-time I ssues in Wireless Sensor Networks

David SI MPLOT-RYL

I RCI CA/ LI FL, CNRS UMR 8022, Université de Lille 1 I NRI A Futurs, France

http://www.lifl.fr/~simplot simplot@lifl.fr

SIMPLOT-RYL – Sensor networks

ETR05

Plan

• 1. I ntroduction (10 min.)
• 2. MAC layer (10 min.)
• 3. Network layer (15 min.)
• 4. Other issues (8 min.)
• 5. Conclusion (2 min.)

SIMPLOT-RYL – Sensor networks

ETR05

Introduction

Sensor networks [Estrin, Mobicom 1999]

Information gathering and processing Data centric: data is requested based on certain attributes Application specific Energy constraint Data aggregation (also data fusion)

Military applications:

(4C’s) Command, control, communications, computing Intelligence, surveillance, reconnaissance Targeting systems

Health care

Monitor patients Assist disabled patients

Commercial applications

Managing inventory Monitoring product quality Monitoring disaster areas

SIMPLOT-RYL – Sensor networks

ETR05

Sensor Nets for Search and Rescue

Inactive Sensor

SIMPLOT-RYL – Sensor networks

ETR05

Sensor Nets for Search and Rescue

SIMPLOT-RYL – Sensor networks

ETR05

Sensor Nets for Search and Rescue

Active Sensor

2

SIMPLOT-RYL – Sensor networks

ETR05

Application I n the UC Berkeley Botanical Garden, 50 “micromotes” sensors are dangled like earrings from the branches of 3 redwood trees to monitor their growth.

SIMPLOT-RYL – Sensor networks

ETR05

Introduction

Design factors [Akyildiz et al. 2002]

Fault Tolerance (sustain functionalities) Scalability (hundreds or thousands) Production Cost (now $10, near future $1) Hardware Constraints Network Topology (pre-, post-, and re-deployment) Transmission Media (RF, Infrared, and Optical) Power Consumption (with < 0.5 Ah, 1.2 V)

Passive sensors

RFID Tags that can provide context- aware information

Temperature Seismic wave

Sensor nodes

25 Motes on Damaged sidewall SIMPLOT-RYL – Sensor networks ETR05

Event-driven model

SIMPLOT-RYL – Sensor networks

ETR05

On-demand model

SIMPLOT-RYL – Sensor networks

ETR05

Ad hoc networks vs sensor networks

Sensor networks are simplified versions of ad hoc networks

Mono-application Communication 1-to-all restricted to sink-to-sensors 1-to-1 restricted to sensor-to-sink

Sensor networks need more than ad hoc networks

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 Global application has priority on life of nodes

Sensor networks are not ad hoc networks

It uses same tools and theory but it is different

SIMPLOT-RYL – Sensor networks

ETR05

RFID Tags

Smart labels

Radio Frequency Identification Tag By opposition to bar code which use optical principles

A stronly limited component:

500 times smaller than a classical microprocessor

Chip with a size of some mm² RFID Tag Intel Pentium 4

3

SIMPLOT-RYL – Sensor networks

ETR05

Principle

Typically, RFID Tags car passive components: they have no battery! Tag are powered by electromagnetic field generated by reader

Communication from reader device to vicinity tags: amplitude shift keying (ASK) Communication from tags to reader device: impedance shift keying (ISK)

courtesy Intersoft SIMPLOT-RYL – Sensor networks

ETR05

EAS Application

Electronic Article Surveillance

Once powered, the tag emits The reader listen channel and activate alarm as early as transmission is detected During checkout, the tag is burner out Problem: aliment the tag whatever the tag orientation

SIMPLOT-RYL – Sensor networks

ETR05

Applications

Batch identification

It is the capability to collect information from a set of tags In opposition to optical identification optique Marathon Automatic clocking in Automatic luggage sorting Automatic inventory 50 items in less than one second

SIMPLOT-RYL – Sensor networks

ETR05

More POPS, smaller POPS…

100µm

Courtesy, Alien Technology SIMPLOT-RYL – Sensor networks

ETR05

The MIT Auto-ID Center Vision of “the internet of things”

Courtesy, Auto-ID Center SIMPLOT-RYL – Sensor networks

ETR05

Networking the physical world

RF Tag Networked Tag Readers Savant Control System

4

SIMPLOT-RYL – Sensor networks

ETR05

Real-Time Constraints

I n some applications

Healthcare, fire surveillance, etc.

delay delivery of alarm is a natural constraint

SIMPLOT-RYL – Sensor networks

ETR05

Plan

• 1. I ntroduction
• 2. MAC layer
• 3. Network layer
• 4. Other issues
• 5. Conclusion

SIMPLOT-RYL – Sensor networks

ETR05

Incoming queue

• Outgoing queue

FIFO

• 2. MAC Layer

What is MAC layer?

Generally assimilated to link layer Its role is to schedule packets and to deal with collisions Physical layer Network layer

• SendPacket

Sense WaitACK ReceivePacket Physical Network Data Link Layer 2 Layer 3 Layer 1

OSI MAC Protocol

Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

• One-Hop Delay Management

Basic tricks:

Network layer labels packets with deadline and packet scheduler uses EDF (multi-hop estimation? Cross-layer with higher layers) Introduction of several priority classes with different outgoing queues It introduces internal packet differentiation E.g. alarm packet has priority on control packet

• Internal differentiation
• Collision

No inter-node differentiation

SIMPLOT-RYL – Sensor networks

ETR05

Inter-Node Differentiation

Two distinct families:

Random access protocols / non-deterministic protocols 802.11 WiFi Contention free protocols / deterministic protocols Time Division Multiple Access (TDMA)

SIMPLOT-RYL – Sensor networks

ETR05

Family of Non- Deterministic Protocols

Example of random access protocol: 802.11 History ALOHA CSMA MACA MACAW 802.11 S-MAC 1970 1980 1990 2000 2010 AOB PNAV

5

SIMPLOT-RYL – Sensor networks

ETR05

Collision-Based MAC Protocols

ALOHA: packet radio networks send when ready 18-35% channel utilization CSMA (Carrier Sense Multiple Access): “listen before talk” 50-80% channel utilization Emit message Start End yes Random wait timeout or NACK 26 % 37 % 37 %

ACK ?

Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

Priorities are defined through different inter frame spacing SI FS (Short I nter Frame Spacing)

Highest priority, for ACK,CTS,polling response

PI FS (PCF I FS)

medium priority, for time-bounded service using PCF

DI FS (Distributed Coordination Function I FS)

Lowest priority, for best-effort data

t medium busy SIFS PIFS DIFS DIFS next frame contention direct access if medium is free ≥ DIFS

Interframe Slots

Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

Backoff Procedure

Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

Backoff Procedure (continued)

I nitial contention window size

CW = CWmin (32 slots)

Backoff is randomly chosen

backoff = rand(0..CW)

Each time that a collision occurs

i.e. Backoff=0 for at least two neighbor nodes CW = min( m × CW, CWmax ) (m=2, CWmax=1024)

Each time that a packet is sent successfully

CW = CWmin Variation (linear decrease): CW = max( CWmin, CW-CWmin ) Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

Introduce Priority

For each priority class

DIFS high priority = short DIFS CWmin high priority = small CWmin CWmax high priorty = small CWmax m high priority = small m

Reminder

Initial contention window size CW = CWmin backoff = rand(0..CW) Each time that a collision occurs CW = min(CWmax, m×CW) Each time that a packet is sent successfully CW = CWmin

medium busy SIFS PIFS DIFS DIFS next frame contention

Déjà vu !

SIMPLOT-RYL – Sensor networks

ETR05

Family of Deterministic Protocols

Classical TDMA (like in GSM)

Node A Node B Node C Node D Node E Slot0 Slot1 Slot3 Slot4 Slot5 Slot6 Slot7 Slot0 Slot1

6

SIMPLOT-RYL – Sensor networks

ETR05

TDMA: Pros and Cons

Pro:

Collision free Estimation of delay is easier (see later)

Cons:

Optimal Slot number depends on maximal degree of the graph Time may be wasted in sparse zones

SIMPLOT-RYL – Sensor networks

ETR05

Unit Disk Graph

SIMPLOT-RYL – Sensor networks

ETR05

TDMA: Pros and Cons

Pro:

Collision free Estimation of delay is easier (see later)

Cons:

Optimal Slot number depends on maximal degree of the graph 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

SIMPLOT-RYL – Sensor networks

ETR05

One-Hop Delay Estimation

Measurement

With time synchronization Simple dedicated hand check Or average in exchanged messages Without time synchronization Hand check and delay is assumed to be symmetrical False assumption in general Problem: delay is known to vary with message length

Estimation

Modeling of MAC layer (e.g. M/G/1) Average delay Delay probability distribution Can take message length into account, number of interfering nodes, etc. Problem: ignore software execution (WCET?)

SIMPLOT-RYL – Sensor networks

ETR05

Plan

• 1. I ntroduction
• 2. MAC layer
• 3. Network layer
• 4. Other issues
• 5. Conclusion

SIMPLOT-RYL – Sensor networks

ETR05

• 3. Network Layer

We limit the presentation to unicast routing Three different approaches

Table driven routing protocols On demand routing protocols Position-based routing protocols

7

SIMPLOT-RYL – Sensor networks

ETR05

Table Driven Protocols

For instance: OLSR

Optimized Link State Routing

Basics:

Each node disseminate topological information to the entire network Broadcasting Each node has a vision of the whole network Shortest path for route discovery

Optimizations:

Minimizing number of messages sent in dissemination phase: Instead of blind flooding for dissemination, use of optimized broadcasting: MPR flooding

SIMPLOT-RYL – Sensor networks

ETR05

Forwarding Nodes Selection

MPR : Multipoint Relay Flooding Protocol [Qayyum et al. 2002]

Principle: select the minimal 1-hop neighbor subset which covers 2-hop neighborhood

Number of transmitted messages is divided by two (medium networks)

Courtesy to Laurent Viennot SIMPLOT-RYL – Sensor networks

ETR05

Table Driven Protocols

For instance: OLSR

Optimized Link State Routing

Basics:

Each node disseminate topological information to the entire network Broadcasting Each node has a vision of the whole network Shortest path for route discovery

Optimizations:

Minimizing number of messages sent in dissemination phase: Instead of blind flooding for dissemination, use of optimized broadcasting: MPR flooding Minimizing amount of topological information necessary for path computation MPR relays are shown to be enough to guarantee shortest path preservation

SIMPLOT-RYL – Sensor networks

ETR05

QOLSR and QANS-OLSR

QOLSR and QANS-OLSR are two variations of OLSR in order to take QoS into account

QOLSR [Badis et al. 2003] Modification of computation of MPR Greedy algorithm which selects first neighbors with small delay Cannot guarantee that minimum delay path is discovered Overhead in dissemination QANS-OLSR [Moraru and Simplot-Ryl 2005] MPR is optimal for dissemination no modification Disseminated information is adapted in order to preserve minimum delay paths [NEW] OLSR for delay [Meraihi Naimi and Jacquet 2005] Disseminated information = the entire neighborhood Use delay distribution instead of average delay Message fragmentation is an option to improve this scheme + ???

SIMPLOT-RYL – Sensor networks

ETR05

On-Demand Protocols

For instance, AODV or DSR Demo

SIMPLOT-RYL – Sensor networks

ETR05

Delay in AODV

Simple approach

Delays in flooding messages respect same delays than routed messages first discovered route is the minimum delay route Questionable since message delay depends on message length

Second approach

Include average delays in route request Target choose in received paths the one with smaller delay Problem: most of optimized broadcasting protocols generates only one path

Two solutions:

Route request which minimizes delay “on-the-fly” Multiple route discovery

8

SIMPLOT-RYL – Sensor networks

ETR05

Neighbor Elimination Scheme and Variations

Neighbor elimination scheme (NES) [Stojmenovic et al. 2002]

Also known as “Wait and see” scheme :-) A node which receives a broadcast message for the first time does not forward it immediately but wait to see if somebody else can do the job The wait duration is randomly chosen The node listens messages and marks neighbors reached by other nodes After wait period, if there are unmarked neighbors, the node forwards the message in order to cover these neighbors S A B C S A B C S A B C Wait = 3 Wait = 1 wait = 2 Node A cancels forwarding

SIMPLOT-RYL – Sensor networks

ETR05

NES and variations (2)

Démo

SIMPLOT-RYL – Sensor networks

ETR05

Solutions

NES adaptation:

Wait period is function of average delay between sender node and itself No guarantee that minimal path is discovered

Multiple route discovery

Trade-off between amount of route request sent and number of discovered routes Dedicated broadcast protocol [Hauspie and Simplot 2003] Message randomly change of SN (sequence number) during flooding More SN you more paths Again, no guarantee that minimal path is discovered

Some optimizations are possible by mixing these two approaches and adding delay prediction for non-covered neighbors

SIMPLOT-RYL – Sensor networks

ETR05

Position-Based Routing Protocols

I f positions are available, geographical routing can be used to send data to sinks

Event-driven model

SIMPLOT-RYL – Sensor networks

ETR05

S D A B

Greedy position based localized routing

[Finn 1987] Localized protocol: S knows only position of itself, its neighbors and destination D S forwards to neighbor B closest to D

SIMPLOT-RYL – Sensor networks

ETR05

S A E C D B

Greedy: SABCD vs shortest path SECD

Localized vs. globalized protocol SP Overhead: messages to maintain global information at each node following mobility and/or sleep/active periods changes

9

SIMPLOT-RYL – Sensor networks

ETR05

D An An-1 A3 A2 A1 Greedy is loop-free

Assume A1 closest to D A2 sends to A3 – contradiction, A1 is closer

SIMPLOT-RYL – Sensor networks

ETR05

A B C D E F S A’ Progress based routing ‘84-86.

Random progress [Nelson, Kleinrock]: A, C or F NFP- nearest forward progress [Hou, Li]: C MFR - most forward within radius [Takagi, Kleinrock]: A

Choose closest projection on SD; minimize SA.SD

SIMPLOT-RYL – Sensor networks

ETR05

S D A

Closest direction

DIRectional routing protocol

Basagni, Chlamtac, Syrotiuk, Woodward MOBICOM’98 (DREAM) Ko, Vaidya MOBICOM ’98 (LAR) Kranakis, Singh, Urrutia CCCG’99 (compass routing)

SIMPLOT-RYL – Sensor networks

ETR05

Transmission radius D H G F E DIR is not loop-free !

Greedy and MFR are loop free [Stojmenovic 1998]

SIMPLOT-RYL – Sensor networks

ETR05

Plan

• 1. I ntroduction
• 2. MAC layer
• 3. Network layer
• 4. Other issues
• 5. Conclusion

SIMPLOT-RYL – Sensor networks

ETR05

• 4. Other Issues: Data

Gathering Two different approaches Traditional reverse multicast/broadcast tree with BS as the sink (root) [Chelius 2004] Three-phase protocol: sinks broadcast the interest, and sensor nodes broadcast an advertisement for the available data and wait for a request from the interested nodes. Energy-efficient route [Akyildiz 2002] Maximum total available energy route Minimum energy consumption route Minimum hop route Maximum minimum available energy node route

10

SIMPLOT-RYL – Sensor networks

ETR05

Data Gathering (2) Sample data aggregation protocols SMECN [Li and Halpern, ICC’01] SPIN [Heinzelman et al, MobiCom’99] SAR [Sohrabi, IEEE Pers. Comm., Oct. 2000] Directed Diffusion [Intanagonwiwat et al, MobiCom’00] Linear Chain [Lidsey and Raghavendra, IEEE TPDS, Sept. 2002] LEACH [Heinzelman et al, Hawaii ICSS 2000]

SIMPLOT-RYL – Sensor networks

ETR05

Data Gathering (3) Directed diffusion Sets up gradients for data to flow from source to sink during interest dissemination (initiated from the sink) LEACH Clusters with clusterheads as gathering points; clusterheads are rotated to balance energy consumption

SIMPLOT-RYL – Sensor networks

ETR05

Data Gathering (4) Flooding is used to build gathering trees Building suitable gathering trees is an open question Flooding is a beaconless protocol but energy consuming Data fusion is possible along transmission bottleneck

SIMPLOT-RYL – Sensor networks

ETR05

Sensor Coverage

How well do the sensors observe the physical space

Sensor deployment: random vs. deterministic Sensor coverage: point vs. area Coverage algorithms: centralized, distributed, or localized Sensing & communication range Additional requirements: energy-efficiency and connectivity Objective: maximum network lifetime or minimum number of sensors

Area (point)-dominating set

A small subset of sensor nodes that covers the monitored area (targets) Nodes not belonging to this set do not participate in the monitoring – they sleep

Localized solutions

With and without neighborhood information

SIMPLOT-RYL – Sensor networks

ETR05

Sensor coverage (2)

Avoid blind points

SIMPLOT-RYL – Sensor networks

ETR05

Sensor coverage (3)

Avoid lost of connectivity

11

SIMPLOT-RYL – Sensor networks

ETR05

Area-dominating set

With neighborhood info [Tian and Geoganas, 2002]

Each node knows all its neighbors’ positions. Each node selects a random timeout interval. At timeout, if a node sees that neighbors who have not yet sent any messages together cover its area, it transmits a “withdrawal” and goes to sleep Otherwise, the node remains active but does not transmit any message

With neighborhood info based on Dai and Wu’s Rule k [Carle and Simplot-Ryl, 2004]

Each node knows either 2- or 3-hop neighborhood topology information A node u is fully covered by a subset S of its neighbors iff three conditions hold The subset S is connected. Any neighbor of u is a neighbor of S. All nodes in S have higher priority than u.

SIMPLOT-RYL – Sensor networks

ETR05

Wu & Dai dominating sets

(Simplified version [Carle and Simplot-Ryl 2004]) Each node has a priority (e.g. its I D)

Variations: <degree, ID> <battery, degree, ID> <random, battery, degree, ID> …

A node is not dominant iff

The set of neighbors with higher priority is connected and covers the neighborhood

SIMPLOT-RYL – Sensor networks

ETR05

Connected area dominating sets

[Carle, Simplot-Ryl 2004][Gallais et al. 2005] We change notion of coverage

A node u is covered by a subset A of its neighborhood if the monitoring area of u is covered by nodes of A

Remaining battery is used as priority Timeout is used to inform transmit priority Examples:

SIMPLOT-RYL – Sensor networks

ETR05

Connected area dominating sets (2)

We obtain a Connected Area Dominating Set

SIMPLOT-RYL – Sensor networks

ETR05

Periodical changes of dominant nodes allow to increase lifetime of the network

• Pb. Synchronization!

Connected area dominating sets (3)

SIMPLOT-RYL – Sensor networks

ETR05

Plan

• 1. I ntroduction
• 2. MAC layer
• 3. Network layer
• 4. Other issues
• 5. Conclusion

12

SIMPLOT-RYL – Sensor networks

ETR05

Conclusion

Most of real-time solution in WSN

Ignore software execution time To be integrated in distribution approach WCET vs distribution approach

Activity scheduling increases the difficulty to obtain efficient protocols

Only position-based protocols and on-demand protocols can deal with such feature

To be addressed more precisely

Energy consumption vs real-time Cross-layered approaches Time synchronization

National CNRS Platform on “Sensor and Self-Organized Networks”

http://www.lifl.fr/sensor

SIMPLOT-RYL – Sensor networks

ETR05

Real-time I ssues in Wireless Sensor Networks

David SI MPLOT-RYL

I RCI CA/ LI FL, CNRS UMR 8022, Université de Lille 1,I NRI A Futurs, France

http://www.lifl.fr/~simplot simplot@lifl.fr with the help of I sabelle AUGE-BLUM and Thomas WATTEYNE (INSA Lyon)