A MeasurementBased Model for Estimating Transmission Capacity in a Wireless Mesh Network

1
A Measurement-Based Model for Estimating
Transmission Capacity in a Wireless Mesh Network
Anand Kashyap1, Samrat Ganguly2, Samir
Das1 1Department of Computer Science, SUNY at
Stony Brook, NY and 2NEC Laboratories,
Princeton, NJ http//www.wings.cs.sunysb.edu
Concept
Sender-side Interference
Receiver-side Interference
Motivation Applications like admission Control,
routing protocols, capacity models, channel
assignment need an interference
model. Interference between links is not
binary. Practical and deployable model needed.
Effect Sending capacity is reduced Measured by
Carrier Sense Factor (CSF)
S
S
I
I
Effect Collisions at the receiver Measured by
Successful Delivery Factor (SDF)
Challenges Physical layer is difficult to model,
so is MAC layer for multi-hop networks. Multiple
Interferers Is interference additive?
R
R

• SDFSI 1 Probability of Collision
• SDFSI 1, if CSFSI CSFIS 1
• SDFSI 1, if CSFSI CSFIS gt 1, RSSIRS RSSIRI
  gt ddB
• SDFSI 1 CSFIS DRRI, otherwise

• CSFSI (Transmission capacity of S in presence
  of I)
• CSF depends on quality of link I ? S
• Quality is determined by Received Signal Strength
  (RSSISI) and Delivery Ratio (DRSI )

Idea Create a non-intrusive measurement based
interference model
Capacity Sending capacity of S
Delivery Ratio of S ? R (1 Probability of
collision at R)
Accuracy of Model in predicting capacity of link
S ? R
Problem
5 Mbps
Correlating CSFSI with DRSI
C
2 Mbps
A
E
F
4 Mbps
B
D
Goal Predict capacity of a link (BC), when a set
of links are active (CA, BD, EF) Analytical
models Too complex to capture the Physical and
MAC layer interaction Empirical models
Measurement for all such combinations is
impossible (O(2n))
WINTECH 2006