7/9/2007 AIIT Summer Course - D# 1 Wireless Embedded Systems and Networking Foundations of IP-based Ubiquitous Sensor Networks Self-Organized Multihop

7/9/2007AIIT Summer Course - D# 1

Wireless Embedded Systems and Networking Foundations of IP-based Ubiquitous Sensor Networks

Self-Organized Multihop Routing

David E. CullerUniversity of California, Berkeley

Arch Rock Corp.July 9, 2007


WSN Communication Requirements

• Local neighbor communication (1 to few)• Dissemination (1 to many)• Data Collection and Aggregation (many to 1)• Point-to-point Transfers (1 to 1)

• Reliably over lossy links• At low power

– Idle listening, management, monitoring

• Adapting to changing conditions• With very light memory footprint


Classical View of Routing• Connectivity between nodes defines the network

graph.– Topology formation

• A Routing algorithm determines the sub-graph that is used for communication between nodes.

– Route formation, path selection• Packets are forwarded from source to

destination over the routing subgraph– At each node in the path, determine the recipient of the next

hop• The selection at each hop is made based on the

information at hand– Sender address, current address, destination address,

information in the packet, information on the node.– Table-driven, source based, algorithmic, …– Who knows the route? Do you determine it as you go?


Basic Approaches• Link state

– Nodes shout (send) and listen (receive) to determine neighbor connectivity.

– Each floods this information throughout (Link State Advertisement) so every node has a map (Link state data base) of the network.

– Any node can determine the path or the next hop.– management protocol deals with changes in connectivity– Classic Example: OSPF

• Distance vector– Nodes maintain routing information about “distance” and

“direction” to destinations– Choose next hop by comparing the cost of routing through

neighbors» Cost(dest D, neighbor b) = linkCost(b) + pathCost(b,D)

– Management propagates routing information» Sequence numbers, etc.

– Classic Example: RIP


What’s different in WSN?• There is no a priori network graph

– It is discovered by sending packets and seeing who receives them.

– The link relationship is not binary.» pairs of nodes communicate with some probability that is

determined by many of factors.– It is not static.

• The embedding of the “network” in space is important.

– Need to get information to travel between particular physical places.

– But the “communication range” is not a simple function of distance.

• addressing & naming – Flat EUID? Hierarchical IP? Topologically meaningful?

Spatially meaningful?


Topology Formation• Much of the “paper protocols” define

connectivity graph with unit disk model– Link(A,B) iff dist(A,B) ≤ R

• OK for rough calculations, but not for protocol design

– Nearby nodes may not be able to communicate.– Far away nodes may be able to communicate.– Nodes that communicated in the past may not be able to

communicate in the future. – Nodes may have intermittent communication depending on

external factors.

• Connectivity is determined by communication– If B receives packet reasonably reliably from A, then A B– If A receives packet reasonably reliably from B, then A B– And if both are true, A B

R


Wireless Routing Protocols• Many wireless protocols in the IP context have

been development in the IETF MANET (Mobile Ad Hoc Networking) working group in the context of 802.11 links carrying traditional TCP/IP point-to-point traffic.

– AODV – ad hoc on-demand distance vector– OLSR – Optimized link state Routing– DSDV - Destination Sequenced Distance Vector– DSR – Dynamic Source Routing– TDRPF - Topology Dissemination Based on Reverse-Path

Forwarding

• Assume a fairly “classic” view of connectivity– Naïve radio


Neighbor Communication

0

11


Fundamental Primitive• Transmit to whatever receivers happen to hear it

• This is the fundamental primitive that is buried underneath complex protocols like Bluetooth, but not made available.

• It is what make it possible to build higher level protocols on the link, especially IP.

• To determine connectivity,– Local broadcast– Respond– on-going protocol to estimate quality of the link

» Packet reliability (sequence numbers, acks)• Note 802.15.4 acks only from a specific destination

» RSSI, LQI, …


Simple Address-Free Flooding Protocol

• Root broadcasts a “new” message to local neighborhood

• Each node performs a simple ruleif (“new” incoming msg) then

take local actionretransmit modified msg

• No underlying routing structure required– The connectivity over physical space determines it.


Route-Free “Flood”

0

11


Flooding• Route free dissemination is extremely useful in

its own right– Issue commands– Network-wide discovery– Join, …

• It is also the network primitive that most “ad hoc” protocols used to determine a route

– Flood from source till destination is reached.– Each node records the source of the flood packet

» This is the parent in the “routing tree”– Reverse the links to form the path back


Data Collection

0

112

2

2

22


Multihop Routing

• Upon each transmission, one of the recipients retransmits

– determined by source, by receiver, by …– on the ‘edge of the cell’


The Problems• Flood causes tremendous contention

– Many good links missed because of collistions– Huge amount of noise

• Many links are not symmetric


Flood Dynamics• Experimental Setup

– 13x13 grid of nodes– separation 2ft– flat open surface– Identical length antennas,

pointing vertically upwards.– Fresh batteries on all nodes– Identical orientation of all

nodes– The region was clean of

external noise sources.• Range of signal strength

settings• Log many runs

Ganesan, Krishnamachari, Woo, Culler, Estrin and Wicker, Complex Behavior at Scale: An Experimental Study of Low-Power Wireless Sensor Networks , UCLA Computer Science Technical Report UCLA/CSD-TR 02-0013

http://lecs.cs.ucla.edu/~deepak/PAPERS/empirical.pdf


Final Tree


Factors• Long asymmetric links are common

– Many children• Nodes out of range may have overlapping cells

– hidden terminal effect• Collisions => these nodes hear neither ‘parent’

– become stragglers

• As the tree propagates – folds back on itself– rebounds from the edge– picking up these stragglers.

• Redundancy – Geometric overlap => <41% additional area

Ni, S.Y., Tseng, Y.C., Chen, Y.S., Sheu, J.P.: The broadcast storm problem in a mobile ad hoc network. MobiCom'99

http://www.cs.berkeley.edu/~culler/cs294-f03/papers/bcast-storm.pdf


Robust Tree Building & Dissemination• Grow the tree slowly• Avoid Contention at every transmission• Avoid Redundant Transmissions• Adapt to huge range of density• Make sure it eventually gets to everyone

– Even if they are not listening– Or arrive late

• With very simple mechanism


Trickle – better than flood• Want the communication rate per unit area to be

constant, regardless of the density of nodes– Lots of nodes, transmit infrequently– Few node, transmit more frequently

• Nodes listen before transmitting• Estimate density based on how many nodes you

hear from– Arrival during timer wait extends timer

• If new value is disseminated by others, no need for you to transmit it.

• Increase delay over time so ambient rate approaches zero.

• Shorten delay when new epoch appears.


Collection

• Common use: monitoring– Collection of nodes take periodic samples– Stream data towards a root node

• Builds and maintains trees

• Root announces interest– depth = 0

• Nodes listen• When hear ‘neighbor’ with smaller depth

– start transmitting data to “best” lower neighbor– set own depth to one greater (and include with data)

• Data transmission continuously reinforces & adjusts routes• Aggregation within nodes or within the tree


Building Neighborhoods & Routes• Node transmits to some unknown set• Candidate nbrs are sources of incoming packets• Estimate of inbound link reliability• Occasionally announce inbound link states

– Provides reverse link estimate to outbound neighbors– Basis for cost-based routing

• Cost-based Parent Selection– depth(me) = MIN nbr(me) depth(i)– loss(me) = MIN nbr(me) loss(i)*est(me,i)– trans(me) = MIN nbr(me) trans(i)+etrans(me,i)

• What about nbrs that don’t fit in the table?– FIFO, LRU, Frequency

Taming the Challenges of Reliable Multihop Routing in Sensor Networks, Alec Woo and David Culler, SenSys. 2003.


Pragmatics• Maintain at least two (preferably three) candidate

parents– Link level retransmission and rerouting– Use acks to determine quality of the link– Throw in a new candidate from time to time.

• Do not record list of children– Insufficient memory to build the table.

• Route update message on a trickle schedule


Question• For such distance-vector dags, what pattern of

node movement is most difficult?


Epidemic NW protocols• Goal: scalable, robust communication algorithms that adapt to changing conditions with little state or protocol

• Basic operation: – From time to time (based on local state and what you’ve heard) transmit a packet of

information to whomever hears the message– From time to time, hear a packet which causes an update to local state– Continuous process to ensure full reliability

• Adapt to density, coverage, interference, loss, schedule, …


Robust OTA Programming

• Every byte must (eventually) be correctly received by all nodes!

• Reliable Pipelined Epidemic Distribution of series of pages

– Constrained storage hierarchy» Packet (32 bytes) << RAM (4K) << program (128K) < external

flash (512K)– Lossy links, Critical Contention– Density-aware– Robust to asymmetric links– Dynamic adjustment of advertisements– Minimize set of concurrent data broadcasts– Spatial multiplexing

• Page Advertise, Request/Fix, Xfer– Density-aware suppression and snoop on each

• Packet CRC + Page CRC• 159 Byte memory footprint

flash

…

Maintain

Request

Transmit


OTA Programming•Learn about the environment after deployment

– sensing range, network characteristics, etc.


OTA Programming•Learn about the environment after deployment

– sensing range, network characteristics, etc.


OTA Programming• Learn about the environment after deployment

– sensing data, network characteristics, etc.• Embedded nature of sensor networks• Network scales reaching thousands of nodes• A necessity in debugging and testing cycle


What is Deluge?•A reliable data dissemination protocol for program images over a multihop network.

•Combined with a bootloader (TOSBoot) Network Programming

0101010101010101011010101101010101010101

Program


Data Representation

•Program divided into pages, each consisting of N packets.

101110

110010

010000

101111

000011

Program

Packets

1234 N

•Reduced RAM requirements •Allows for spatial multiplexing


How Does Deluge Work?(Glossing over many details)

•Nodes periodically advertise– Suppress similar advertisements

Version 2 here.

I only have version 1.




•Neighboring nodes request data– Suppress similar requests

Send me page 1!

Send me page 1!



•Requested data is broadcast

Packet 12of page 1!



•Dropped packets are NACKed

Repeat packet 4 of page 1!

Repeat packet 32 of page 1!



•Dropped packets are sent again

Packet 4 of page 1!



•Advertise for propagation to next hop

Version 2 here.



Spatial Multiplexing

•Propagate in “waves”

•Exploit limited radio range for concurrent broadcasts.

•Reduced completion time •o(d + Sobj) vs. o(d * Sobj)

Page 0Page 1


Epidemic Propagation•Epidemic propagation from one source


Epidemic Propagation•Epidemic propagation from one source or many


Deluge Features•Epidemic propagation from one source or many

– Continuous propagation effort by all nodes– Turn on/off radios at will– Reach nodes with intermittent connectivity

Will find a path if it exists

•Aggressive message suppression– Scales with density– Ultra low quiescent traffic


Deluge Features• Management

– Multiple program images– Image metadata– User confirmation on expensive operations

»Minimize operator error

• Robustness– Redundant CRCs– Golden Image with write protect– Load Golden Image

»Watchdog trigger»Golden gesture

• TOSBoot– TOSBoot as isolated code– Verify CRCs– Verify system voltage

0101010101010101011010101101010101010101

0101010101010101011010101101010101010101

0101010101010101011010101101010101010101

ProgA ProgB ProgC

Program NameCompile TimeUserIDHostnamePlatform

CRASH!

CR

CC

RC

CR

CC

RC

CR

CC

RC

CR

CC

RC

CR

CC

RC

CR

CC

RC01010

10101010101011010101101010101010101


Deluge - Lessons Learned•Advantages

– Ease of reprogramming 100’s-1000’s of nodes– Does not erase node IDs– Golden Image is immensely useful– Quickly switch between images– More reliable than uisp or msp430-bsl– Deluge over 802.15.4 more efficient that 802.11!

•Disadvantages– Ease of reprogramming 100’s-1000’s of nodes


Integrated, Density-Aware Dissemination and Collection

Drip – packet dissemination layer

GW

• Trickle neighborhood suppression lowers bandwidth requirements

• Trickle exponential retransmission creates epidemic reliability

• Provides “transport-layer” interface

Drain – packet collection layer

GW

• 802.15.4 LQI for link estimation• Gateway initiates the tree-build• Fast, minimal RAM, evolvable• Reliability from link-layer ACKs and

exponential retry with long window

PC

802.

11

Documents

7/9/2007 AIIT Summer Course - D# 1 Wireless Embedded Systems and Networking Foundations of IP-based Ubiquitous Sensor Networks Self-Organized Multihop