an and d Im Impl plementa ementation tion (01 0120 20442 - - PowerPoint PPT Presentation

Net etwork work Ke Kerne nel l Ar Archi chitect tectures ures an and d Im Impl plementa ementation tion (01 0120 20442 4423) ) Net etwork work Ar Archi chite tecture cture

Chaiporn Jaikaeo chaiporn.j@ku.ac.th Department of Computer Engineering Kasetsart University

Materials taken from lecture slides by Karl and Willig

2

Out Outline line



Ne Netwo twork rk sc scena nari rios

• s



Optimization goals



Design principles



Gateway concepts

3

Ty Typi pical cal Vi Views ws of f WS WSN



Self-organizing mobile ad hoc networks (MANETs)



Peer-to-peer networks



Multi/mobile agent systems and swarm intellegence

4

Sens nsor

• r Ne

Netw twork rk Sce cena narios rios



Sour urces ces: Any entity that provides data/measurements



Sin inks: Nodes where information is required

Source Sink Internet Sink Source Sink Source

5

Sin ingle gle-Hop Hop vs. . Mul ulti ti-hop hop



Multi-hop networks

• Send packets to an intermediate node
• Intermediate node forwards packet to its destination
• Store

re-and and-fo forward rward multi-hop network



Store & forward multi-hopping NOT the only possible solution

• E.g., collaborative

networking, network coding

Source Sink Obstacle

6

Mu Multi lti-hopping hopping Al Alwa ways ys Ef Effi fici cient? nt?



Obvious idea: Multi-hopping is more energy- efficient than direct communication

• Suppose we put a relay at distance d/2
• Energy for distance d is reduced from cd to

2c(d/2)

• c - some constant
•  - path loss coefficient ( 2)



Usually wrong, or over-simplified

• Need to take constant offsets for powering

transmitter, receiver into account

7

Mu Multiple ltiple Sin inks, ks, Mu Mult ltiple iple sou

• urces

rces

8

Out Outline line



Network scenarios



Optimization timization go goals ls



Design principles



Gateway concepts

9

Go Goal: al: Qua Quali lity ty of f Serv rvic ice



QoS in WSN is more complicated (compared to MANET)

• Event detection/reporting probability
• Event classification error, detection delay
• Probability of missing a periodic report
• Approximation accuracy (e.g, when WSN constructs a

temperature map)

• Tracking accuracy (e.g., difference between true and

conjectured position of the pink elephant)



Related goal: robustness

• Network should withstand failure of some nodes

10

Go Goal: al: En Energ rgy y effi ffici ciency ncy



Many definitions

• Energy per correctly received bit
• Energy per reported (unique) event
• Delay/energy tradeoffs
• Network lifetime
• Time to first node failure
• Network half-life (how long until 50% of the nodes

died?)

• Time to partition
• Time to loss of coverage
• Time to failure of first event notification

11

Sha harpe rpening ning th the Dr Drop



Sacrifice long lifetimes in return for an improvement in short lifetimes

12

Out Outline line



Network scenarios



Optimization goals



Desi sign gn prin inciples iples



Gateway concepts

13

Di Distr tributed ibuted Or Organ ganization ization



WSN participants should cooperate in

• rganizing the network
• Centralized approach usually not feasible



Potential shortcomings

• Not clear whether distributed or centralized
• rganization achieves better energy efficiency



Option: “limited centralized” solution

• Elect nodes for local coordination/control
• Perhaps rotate this function over time

14

In In-Ne Network twork Pro roce cessing ssing



WSNs are expected to provide information

• Gives additional options
• E.g., manipu

pulate late or proc

• cess the data in the

network



Main example: aggregation

• Apply aggregation functions to a collection tree

in a network

• Typical functions: minimum, maximum,

average, sum, …

• Not amenable functions: median

15

Ag Aggr gregation gation Ex Exam ample ple

1 1 3 1 1 6 1 1 1 1 1 1

16

Sig ignal nal Pro rocess cessing ing



Another form of in-network processing



E.g.,

• Edge detection
• Tracking/angle detection of signal source



Exploit temporal poral and sp spatial tial correlation elation

• Observed signals might vary only slowly in time
• Signals of neighboring nodes are often quite

similar



Compressive sensing

17

Ada Adapt ptiv ive e Fi Fide delity lity



Adapt data processing effort based on required accuracy/fidelity



E.g., event detection

• When event occurs, increase rate of message

exchanges



E.g., temperature

• When temperature is in acceptable range, only

send temperature values at low resolution

• When temperature becomes high, increase

resolution and thus message length

18

Da Data ta Cent ntric ric Ne Netw twork rking ing



Interactions in typical networks are addressed to the id ident ntit ities ies of nodes

• Known as node-centric or address-centric

networking paradigm



In WSN, specific source of events might not be important

• Several nodes can observe the same area



Focus on data/results instead

 Data-centri centric c networ

• rking

ng

• Principal design change

19

Im Impl plementa ementation tion Opt Optio ions ns



Publish/subscribe (NDN – Named Data Networking)

• Nodes can publ

blish data, can subs bscr cribe be to any particular kind of data

• Once data of a certain type has been

published, it is delivered to all subscribers



Databases

• SQL-based request

20

Out Outline line



Network scenarios



Optimization goals



Design principles



Gat ateway eway con

• ncep

epts ts

21

Ga Gate teways ways in in WS WSN/ N/MANET MANET



Allow remote access to/from the WSN



Bridge the gap between different interaction semantics

• E.g., data vs. address-centric networking



Need support for different radios/protocols

22

Gateway nodes Internet Gateway

WS WSN N tu tunne nneling ling



Use the Internet to “tunnel” WSN packets between two remote WSNs

6Lo LoWPAN WPAN



IPv6 over Lo Low-power Wireless Personal Area Networks



Nodes communicate using IPv6 packets



An IPv6 packet is carried in the payload of IEEE 802.15.4 data frames

23

Ex Example ample 6Lo LoWPAN WPAN Sys yste tems ms

24

25

Sum ummary mary



Network architectures for WSNs look quite different from typical networks in many aspects



Data-centric paradigm opens new possibilities for protocol design