Routing in Sensor Networks Prabal Dutta CS 294-11, Oct 25, 2005 Some Communication Abstractions  Collection (MintRoute)  Dissemination (Trickle)  Point-to-Point (BVR)  Aggregation (TAG,

Download Report

Transcript Routing in Sensor Networks Prabal Dutta CS 294-11, Oct 25, 2005 Some Communication Abstractions  Collection (MintRoute)  Dissemination (Trickle)  Point-to-Point (BVR)  Aggregation (TAG,

Routing in Sensor Networks
Prabal Dutta
CS 294-11, Oct 25, 2005
Some Communication Abstractions
 Collection (MintRoute)
 Dissemination (Trickle)
 Point-to-Point (BVR)
 Aggregation (TAG, Synopsis Diffusion)
 Neighborhoods (Hood)
 Data-centric Storage (GEM, PathDCS)
 Attribute-based Routing (Directed Diffusion)
Slides borrowed from:
A Holistic Approach to
Multihop Routing for Sensor Networks
Alec Woo
Dissertation Talk
Computer Science Division, UC Berkeley
with David Culler and Terence Tong
Key Takeaways
Physical connectivity is not unit disk
What does connectivity look like?
How to estimate connectivity?
Often, more neighbors than slots in NBR TBL
When to insert? Evict?
How to avoid thrashing?
Routing algorithms use cost metrics
What are the right metrics? Hops? Distance? METX?
Collection routing is a very common pattern
Boolean Connectivity Assumption
2
2
2
2
1
1
2
0
A
Measure
Physical Connectivity
Average link quality among many pairs of nodes at
different distances
Communication Range?
3 regions, with a large transitional region
Effective
Transitional
Region
Region
Clear
Region
Implications
Transitional
Region
Deployment: (X-axis) (In-situ analysis)
Communication range = effective region
Individual nodes (Y-axis)
Discover connectivity = link estimation
Hear many nodes in transitional region
How to define a “neighbor”?
Zhao et al., SCALE
Neighborhood: A Fuzzy Concept
 Many potential neighbors
 Short effective region
 Short sensing range
 Few good ones (blue)
 Large gray region
 Neighbors > Table-size
Get in
Neighbor
Table
Get out
 If not in table,
 can’t estimate
 Don’t rely on density
control
 Adapts to all cell density
General solution:
down-sample to suppress
gray nodes
maintain frequent nodes
Average Hop-Count Contour Plot
Derive Connectivity Graph through
Passive Link Estimation
 Link sequence number snooping
 Estimate inbound reception quality
 Key issue
 Cannot infer losses until next packet reception
 Solution
 Rely on a network-wide minimum data rate
 infer losses based on it
 Bi-directional estimation
 Require outbound transmission quality estimation
 Exchange reception quality over local broadcast
 E.g piggyback on route updates
A Good Estimator
Accurate
+/- 10% error, with a high confidence
Agile yet stable
Relative to message opportunities rather than time
Small memory footprint
Many neighbors to estimate!
Simple
This is a low-level operation
On-Line Table Management Process
Insertion Policy
Adaptive down-sampling hysteresis
Throw a coin, only insert if success
Eviction and Replacement Policy
Classical Cache Replacement Policy
FIFO, LRU (LRH), Clock
Borrow Database Techniques
Estimate most frequent tokens of a data stream
FREQUENCY (Manku et al.)
Key Results
Fixed-size table as cell density increases
1st
2nd
3rd
# Good
neighbors
> Table size
Freq always
keeps 50%
or more of the
table entries
in maintaining
the good
neighbors
40
Number of Potential Neighbors
Cost Functions
SP on physical connectivity graph
SP with threshold on logical connectivity graph
Path Reliability (Yarvis et al.)
Product of link quality along the entire path
Exponential drop: (link success rate)# of hops
Assumes no link retransmissions
Minimum Transmission (MT)
Cost is based on link quality
1
Link estimator 
hop p forward  preverse
provides
Cost = E[total number of trans.]
 ETX (De Couto et al.)
Implicit retransmission assumption
70%
70%
50%
Tree-Building Approach
 Variant of a distributed distance-vector protocol
 Goal: stable and reliable tree (nodes are relatively immobile)
 Different from discovering paths quickly in mobile computing
 Operate over a dynamically changing physical connectivity graph
 Environmental changes
 Node failures
 Low-rate periodic route messages (low bandwidth)
 Carry “cost” to tree root
 Piggyback link estimations
 Hear neighbor’s “cost” and store in table
 Select minimum cost neighbor for routing
 Route damping (stability)
 Periodic vs. asynchronous
 Switching threshold for noisy cost
Self-Organizing Networks
 Using only simple local rules for highly
resource-constrained nodes to self-organize into
a globally consistent and robust network
Protocol design consideration
Bandwidth/energy
Amount of states/complexity
Memory footprint
One instance: Multihop routing
Overview
 Problem decomposition into 3 local processes
 Connectivity defines relative to link quality estimation
 Neighbor table management to build weighted logical
connectivity graph
 Cost functions to exploit such graph
 Observe global properties
 End-to-end success rate
 Hop distribution
 Topology Stability
 Extensive simulations and empirical experiments
 MintRoute, released in TinyOS 1.1
Roadmap
Physical Connectivity in Reality
Connectivity Graph Derivation with Link
Estimations
Neighborhood Management
Tree-Based Routing Study
Central Limit Theorem Prediction
For a 10% error with a 95% interval
 worst case for agility is at least 100 packets
2
[
]  10% 95% ConfidenceInt erval
n
  p (1  p ) Sigma for BinomialDist.
4
n
  p (1  p )
2
(10%)
n  100
Wosrt Case : p  0.5
Estimator Study
 Study 7 different estimators
 EWMA, Flip-Flop EWMA, MA, Time-weighted MA, Packet
Loss/Success Interval, WMEWMA
 Compared by tuning each to the same objectives
 Verify with empirical traces
 See details in thesis
 Results
 WMEWMA(T, ) Estimator
 Stable, simple, constant memory footprint
 Compute success rate over non-overlapping window (T)
 Average over an EWMA()
 Key Implication
 10% |error| requires at least 100 packets to settle
 Limits rate of adaptation
Roadmap
Physical Connectivity in Reality
Connectivity Graph Derivation with Link
Estimations
Neighborhood Management
Tree-Based Routing Study
Details
 Insert
 Set prob. such that insertion rate < reinforcement rate
 Down-sample prob.  min(1,Table Size / # Neighbors Est.)
 Estimate # neighbors based on periodic route beacons
 Reinforce if in table
 Cache hit (FIFO, LRH, Clock)
 Node’s Counter++ (Freq)
 bypass down-sampling for reinforcement
 Evict
 Cache policies
 evict for each insertion
 Freq: Counter--,
 Counter == 0 becomes replaceable
 If all Counters > 0, drop insertion
Implications
 Non-threshold based neighborhood selection
 No estimation required
 One-hop neighbor
 Based on competitiveness relative to the goodness metric
 Other goodness metric that augment neighborhood
selection
 Control in/out degree on the logical connectivity graph
 Higher-level changes on cell density will not affect
system functionality
 Connectivity graph adapts with its best using limited resources
 New neighborhood interface and abstraction
Holistic Approach to Routing
Now, the connectivity graph is built
A
A
A
Select Good Routes
Based on ?
Neighbor management
using FREQUENCY
Link Estimation using
WMEWMA
Many-to-One Data Collection
A common routing service for data collection
Simple form of directed-diffusion
Tree rooted at the sink node where data is collected
Single hop weather
Single hop burrow
Multi hop weather
Multi hop burrow
10m
Evaluation Roadmap
 Key observations:
 Hop distribution, end-to-end success, stability
 Graph analysis
 80x80 grid
Large
 SP, SP(%), MT
 Rule out SP because of poor reliability
High
Level
 Packet-level simulation
 10x10 grid, (max 2 retrans./hop)
 Broadcast and DSDV (periodic route selection)
 Neighbor table management
 Freq + Routing Goodness -> MTTM
 Empirical (Mica/Mica2 Motes)
 5x10 grid and 30-node random placement, smote
 SP(%), MT with large enough table
Small
 max 2 retrans./hop, deliberate congestion
Low
Level
Graph Analysis Key Results
Hop-Distribution and Reliability to BS
Simulation Key Results
Stability
Hop-Count
Distribution
End-to-end
Success
vs. Distance
Empirical Study
Restudy connectivity vs. distance
Put nodes at end of effective region (~ worst case)
8 feet
Study SP(70%), SP(40%), MT
Link
Key observations: quality
SP(70%) fails
drops
under
traffic
SP(40%) fails
Hard threshold fails
under congestion
Empirical Key Results
Hop-Count
Distribution
End-to-end
Success
vs. Distance
Effective Region is 8 feet
Different
from
simulations!
Congestion and Stability
Topology
Stability
# Route Changes
Per 5 Route Messages
30-node network
Link
%
Estimation
Time (s)
Possible Congestion/Rate Control: Woo et al. (Mobicom ’01)
Mitigate Instability
Subtle overflow bug in link estimation
Confidence-interval filtering on link estimation
Link estimation to tree root can affect stability
on the entire tree
Switching threshold helps stability, but
sacrifices end-to-end success rate
Cross-layer Interactions
Ave.
# of
Parent
Changes
Per
Route
Update
3.02
2.49
0.52
0.14
0.10
Induced Interference
Ave.
# of
Parent
Changes
Per
Route
Update
0.30
0.10
Node Failure
Current Status
Used by GDI ’03, TinyDB, TASK (Intel)
TinyOS 1.1 Release
Surge as a Network Analysis Tool
Crossbow: www.xbow.com
Incorporated with
low-power listening
~97% success rate on mica2
Source: Crossbow
Related Work Summary
 Connectivity Study
 Choi et al., Zhao et al., Cerpa et al., Ganesan et al.
 Link estimation
 IGRP, EIGRP, De Couto (Mobicom ’03), Kim et al. (Mobicom ’99)
 Neighborhood Management
 Limiting Logical Neighborhood Size (Miller et al., Simulation of
computer networks ’ 87)
 Random Selection (Shacham et al., ICC ’88)
 Routing Metrics
 De Couto (Mobicom ’03)
 Draves et al. (Microsoft Research TR-2004-18 March ’04)
 LIR, least gain routing opt. for spatial reuse (SRNTN ’88)
 LRR, link cost = physical-level interference, (Tactical
Communication Conference ’90)
 Sensor Network Routing
 Real experiment running DSDV + Path Reliability Metric (Yarvis
et al. IWAHN ’02)
Future Work
Reverse Tree Routing Support
 any-to-any routing
Co-design of query processing and networking
Query-informed routing
See June Communication of the ACM ‘04
Thank you!
Backup Slides
A Connectivity Cell
144-node, 12x12 grid network with Rene Motes
Joint work with
Ganesan et al.
2-feet spacing
Low transmit
power
Open tennis
court
RSSI & Link Quality
 Can we use RSSI to predict link quality?
Low packet loss => good RSSI
But not vice versa
Interference from traffic
Similar findings
Zhao et al. (RFM sensor networks)
De Couto et al. (802.11 networks)
Approximate
Connectivity Variations
Approximate time variations
Time-Varying Connectivity
Link quality varies over time
over a 5-hour period
over an 8-hour period
Routing Architecture
Send originated data message
Send route update message
Cycle detected
• choose other parent
Originating Queue
Timer
Application
Run parent selection
and send route
message periodically
Parent
Selection
Forward Queue
Cycle
Detection
Forwarding
message
Filter
• discard non data packet
• discard duplicate packet
Data
message
Estimator
All message
• sniff and
estimate
Neighbor
Table
Table
Management
Route message
• save information
All Messages
Topology over Time
Est. Link Quality
70-100%
49
40-70%
42
0- 40%
Feet
35
Tree Depth
28
1
21
2
14
3
7
0
7
14
21
28 35
Feet
42
49
56
63
Channel Utilization Contour
Routing Cost: Actual vs. Est.
Pursuer and Evader Application
 “Design and Implementation of a Sensor Network System for
Vehicle Tracking and Autonomous Interception”, Submitted to OSDI
2004
The Berkeley NEST team
Hops and Cost Metrics
Shortest Path vs. Shortest Path with threshold
Hop over distance is a relative concept.
Highlights of Other Work
 Query Processing and Networking Co-design
 CACM June 04, with Ramesh Godvidan and Sam Madden
 Shadowing Phenomenon
 UCB Tech 04, with Kamin Whitehouse, Joe Polastre, Fred Jiang
 Ranging and Localization
 Acoustic, Ultrasound
 Infrastructure and Ad hoc
 Submitted to SenSys 04, with Kamin Whitehouse, Fred Jiang,
Chris Karlof, and David Culler
 Mica Sensorboard
 Sold as Crossbow MTS300/310
 MAC and Transmission Rate Control for Fairness
 Mobicom 2001, with David Culler
 TinyOS
 ASPLOS 2000
 with Jason Hill, Robert Szewczyk, Seth Hollar, David Culler, and Kris
Pister
2004: a year of the mote?
May be?
What can you really do with it?
“I think there is a world market for maybe five
computers.”
- IBM Chairman Thomas Watson, 1943
It’s time to innovate! Let’s talk!
Why such a Holistic Approach?
The underlying issues matter!
Expose and embrace these issues
Not assume over them
Articulate the 3 core system components
Understand how they interact and affect each other
Independent improvement
Reusability
Distributed Tree-Building Process
Candidate Non-Bayesian
Link Estimators
Select Good Routes
Over Logical Conn Graph
Neighbor management
 keep the good ones
 build a logical
connectivity graph
A Derived
Connectivity Graph
Wireless Networking
Rooftop/Metropolitan Networks
Packet Radio Networks
Wi-Fi Mobile Computing
Sensor Networks
Application
s
Traffic
global
Individual User Network as a whole
Co-op, correlated,
Pairs of indep.
flows(end-to-end) in-network processing
Transport
End-to-End
Routing
local Any-to-any
Mobility
Mobile
Resources
Not a concern
Radio
High
Bandwidth
Phy Layer
Spread spectrum
?? Custody/Best Effort
Many-to-one(few)
Static
Limited
Low
Single-band
Challenges
Programming a large network of highly resourceconstrained nodes to self-organize into some
global consistent and robust behavior using
only simple local rules over a noisy and
dynamically changing environment
Think small and big
Take a probabilistic view to describe lossy link quality
and follows such apporach all the way up to the
routing layer
Bandwidth/energy, amount of states/complexity,
memory footprint, reliability over unreliable channel
2004: a year of the mote?
May be?
I think there is a world market for maybe five
computers (sensor networks?).
- IBM Chairman Thomas Watson, 1943
There is no reason anyone would want a
computer (sensor network?) in their home.
-Ken Olson, president of Digital Equipment Corp. 1977