4. Characteristics and Protocol Mapping
Each layer within the protocol stack is defined by a number of characteristics.
For example, routing sub-protocols can be classified using three orthogonal
characteristics, or three ‘alphabet’ letters and can be modelled as a 3D vector
space. Each protocol can be mapped to a point in the vector space defined by a
Vector space of the routing protocol classification for Qualnet
scale along each characteristic.
Update Period
(2) Hybrid
(3) Periodic
Source Routing
(1) Source
OSI Model
2
1.5
1
3
(2) Hybrid
(3) Hop
Characteristic scale
2
1
Simulation of the
protocol in an
ad hoc network
QualNet
Alphabet
Performance Measurements
3.5
5000
3
4000
Genetic Algorithm
Genetic algorithm
to generate ‘ideal’
protocol
Generate initial
population of N
random protocols
Matlab
2000
2
1000
1.5
0
10
20
30
40
generation number
50
60
ROUTING PROTOCOL
Terminate
?
4
source routing (H)
Determine fitness
of each protocol
The full simulation environment uses Matlab to
Run a genetic algorithm to find the ‘ideal’ protocol, and a
simulator to simulate the ad hoc network environment
and extract performance measurements.
0
7.0Optimum Protocol
0
1
2
3 Quality of
4 Service (D
Optimisation of the
routing protocol under a
Range Type (A)
particular network scenario found AODV (Ad
hoc On-demand Distance Vector) to be the
fittest.
Ad hoc network
Ad hoc network
Ad hoc network
simulation
simulation
simulation
Selection,
Crossover &
Mutation
Fault injected
into network
2.5
3000
0.5
Generate Protocol
Generated protocol
6000
1
Protocol
Characteristics
Available
sub-protocols
4
route computation (c1)
A genetic algorithm is used to identify a point in the vector space that is
considered to be the best guess of the routing protocol characteristics under the
current network conditions. This is then mapped to the closest available routing
protocol . Each layer of the protocol is built up in this way.
5. Simulation
Environment
7000
2
1
MA
Mean Fitness Score
over Generations
PHYSICAL
PROTOCOL
3
update period (c2)
Test Protocol
3. Protocol Stack
Communicating between devices involves
layers of interacting processes, from the
transmitting radio waves to the user software
such as a web page. These combined layers are
called a protocol stack. Applications Application
And
Layer
A standard model of
Middleware Presentation
a protocol stack is the
Layer
OSI model. Each layer
Session
Layer
can contain one or
Networking Transport
more different subLayer
protocols. Protocols
Network
today generally use a
Layer
subset of these subEnabling
Data Link control
MAC
Technologies Layer
protocols and are
Physical
fixed for a particular
Layer
type of application.
2.5
Number of channels (E)
(3) Proactive
fitness value
(2) Hybrid
3
6. Genetic Algorithm
A genetic algorithm is used to find an optimum
solution by minimising a simple fitness
function. An initial population of N random
protocols are simulated in turn, each returning
performance measurements, then processed by
the fitness function to obtain a ‘fitness score’.
Selected ‘fittest’ protocols undergo crossover
and mutation to create a new population. This
is repeated until an optimum solution is found.
Max Bit Rate (B)
(1) Reactive
(1) Event
2. Future of Ad hoc Networks
An ad hoc network is characterised
by devices connected in an arbitrary
manner to form a network without a central
controller. Ad hoc networks of the future must
interconnect applications such as home robots,
wearable computers, and sensor networks. To
succeed they must be self-organised and display
characteristics of emergence, robustness,
adaptability and scalability.
LSR
FSR
DSR
AODV
ZRP
STAR
Route Computation
source routing(c3)
1. Research Objective
The objective is to produce a protocol for ad
hoc networks of the future via emergence. Such
a protocol needs to accommodate scenarios
which can not be premeditated prior to
deployment. This involves the creation of an
‘alphabet’ from the characteristics of existing
protocols with methods
for combining
the letters in it.
3
2
1
0
4
4
Best chosen
protocol
2
update period (G)
0
0
2
route computation (F)