Stability-aware Cognitive Packet Network Routing Protocol for

MANET

A. Alharbi

, A. Aldhalaan

and M. Alrodhaan

Department of Computer Science, Princess Nora University, Riyadh, Saudi Arabia

Department of Computer Science, King Saud University, Riyadh, Saudi Arabia

1 STAGE OF THE RESEARCH

The PhD research is currently in the simulation

phase of the routing protocol using OPNET 14.5

modeler. There are three levels of simulation

modelling in OPNET. First, there is the OPNET

Network model, in which the research defines the

MANET to perform the different simulation

scenarios upon. Then there is the Node model,

where the Cognitive Node is defined in detail as

shown in section 3. The last level is the Process

model, where the details of the routing protocol is

defined as a process state diagram.

The research is in the proess model stage,in

which the routing protocol states, events, conditions,

and actions are defined. First the INITIAL state is

identified, then more states are added as needed. The

process model is then saved in the OPNET Process

Editor. Finally, program code segments are added in

its appropiate places in order to run the simulation of

the protocol.

2 OUTLINE OF OBJECTIVES

The Cognitive Packet Network (CPN) Routing

Protocol is a unique adaptive protocol with self-

improvements capabilites (Gelenbe, 2011). It offers

the network the ability to determine the Quality of

Service (QoS) criteria according to the data being

transferred (application) at the software level in a

distributed way. Each node in the network runs the

protocol using Computational Intelligence to find

the suitable path for every packet. Network

information is collected only for paths being used;

there is no global network information exchange.

This information is used by a Reinforcement

Learning Algorithm and a Random Neural Network

to take routing decisions at every node according to

specific Goal Function depending on the needed

QoS criteria.

This research extends the work on CPN Routing

Algorithm to adapt the protocol to the Mobile Ad

hoc Network (MANET) environment. MANETs are

wireless networks where packet communication

does not depend on any infrastructure. Thus a

MANET is characterized by highly unstable

topology. Node mobility is a major problem and so

links between nodes are broken frequently. The

research focuses on node stability within its

neighbours as criteria for choosing a node as part of

a routing path.

The research routing algorithm defines a Goal

Function of a combination of high-stability and

short-delay criteria. The nodes that satisfy this goal

are chosen for routing with high probability. Thus

the algorithm results in long-live, short paths while

being fast adaptive to topology changes.

The research also shows that using

Computational Intelligence in a challenging routing

environment such as MANET gives comparable

results to conventional MANET routing protocols

without disrupting the overall packet delay.

3 RESEARCH PROBLEM

CPN performs routing using three types of packets:

smart packets (SP), dumb packets (DP), and

acknowledgments (ACK). SP is used for route

discovery and route refinement and maintenance. DP

carry the actual data. ACK carry feedback

information about the routes discovered. All packets

have the same structure: a header, a cognitive map.

and the payload data. A cognitive map holds

information about the nodes visited by the packet

and the visiting time. To discover routes SP’s are

source-initiated to move through the network

gathering specific network information according to

the specific QoS goals determined in each SP. Until

the requested path is found SP’s are continuously

sent by the source node. Once the first ACK reaches

the source node carrying the first path discovered,

the rate at which the SP’s are sent is reduced in order

to maintain and improve QoS delivered.

Alharbi A., Aldhalaan A. and Alrodhaan M. (2013).

Stability-aware Cognitive Packet Network Routing Protocol for MANET.

In Doctoral Consortium, pages 3-7

 SCITEPRESS

At each node, when a SP arrives the node checks

if the SP is a duplicate and thus discarded. If the SP

is not a duplicate, the node checks if itself is the

destination node. If the node is the destination, it

creates an ACK packet for this SP. The ACK uses

the reverse route of the route discovered by the SP.

The ACK passes every node on the discovered route

and updates the weights in RNN according to the

performance.

On the other hand, if the node is not the

destination, then it should use the RNN to make a

decision on which outgoing link to send the SP when

enough neighbour information is available. However

if the neighbour information is not sufficient to

make a decision, then the node broadcasts the SP to

all of its neighbours on all outgoing links.

As soon as the first ACK reaches the source

node, the source node copies the discovered route

into all DP’s ready to be sent to carry the payload

from source to destination. DP’s use source routing

with the discovered route until a new ACK brings a

new better route to the source node. Thus SP’s

continue to be issued from the source node at a low

rate to maintain and improve QoS.

The CPN algorithm uses the Goal Function of a

combination of low power consumption and short

delay (QoS) to make routing decisions on wired

networks. The research algorithm introduces the

node stability to the goal function of CPN in order to

adapt it to the MANET environment. Such that

nodes with high stability are chosen for routing with

high probability. In this algorithm stability is

measured through the node's neighbour associativity

degree over time and space (Toh, 2004).

Each node sends out low frequency beacons to

signify its existence. It also keeps track of all its

neighbours and the number of beacons it received

from each neighbour in a table. If the number of

beacons received from a certain neighbour is more

than a specified threshold, the neighbour is

considered stable and could be used in the routing

path. The associativity threshold is a function of the

beaconing interval (p), the relative velocity between

the two nodes (v), and the transmission range (r) of a

node as shown in equation 1 (Toh, 2000).

= 2r/ pv (1)

Associativity degree is reset when either the mobile

node itself or the neighbour move out of

transmission range. The associativity property

assumes that a mobile node goes through a stage of

instability with high mobility followed by a stage of

stability when it is dormant (connected to the same

neighbours for some time) before the mobile node

moves out of proximity. The dormant stage is the

best time for a node to participate in routing, which

is determined by a high associativity level.

The research routing algorithm uses a goal

function which optimizes the nodes stability while

looking for short routes. It aims to find long-live

routes without disrupting the delay constraint. Thus

for some active flow k, the algorithm computes the

routing goal at node i (G ) using equation 2.



















(2)

where 





is the total node associativity level of all

path nodes from node i to the destination node. And







is the total path delay for flow k measured

from node i to destination node. At each node, a

separate Random Neural Network (RNN) is stored

for each flow k. Each neuron in RNN is associated

with a specific output link for that node. When a SP

is received the RNN is used to make a routing

decision to send the SP on one output link.

When an ACK is received, the Reinforcement

Learning algorithm is used to reward or punish the

chosen path according to the delivered QoS. The

research algorithm defines a reward function as

shown in equation (3).

R = 1/ G (3)

Successive measurements of the reward are

collected and RNN weights are updated according to

this historical average threshold. First the threshold

is computed as shown in equation (4).

(4)

Then the current reward R

is compared to the

previous threshold T

l-1

. The excitatory weights of all

the neurons up to that neuron are significantly

increased if the current reward is larger than the

previous threshold, with slight increase in the

inhibitory weights which leads to other neurons.

However if the current reward is less than T

meaning that the chosen neuron was not successful

of producing better reward, then all the inhibitory

weights leading to the winner neuron are

significantly increased. While all excitatory weights

of other neurons are increased moderately. The

Reinforcement Learning algorithm (Gelenbe and

Seref, 2001) performs this weight update as shown

in equation (5).

IJCCI2013-DoctoralConsortium

(5)

where i, j, and k are the output links to the node's

neighbours other than the link which the packet

came from. And j is the output link that was most

recently used to route packets. Also in equation (4) n

denotes the number of neighbours to the node

currently updating its RNN weights.

4 STATE OF THE ART

The Stability-Aware CPN Routing protocol for

MANET is a unique research protocol which

introduces node stability over space and time into

the CPN Ad hoc extension routing protocol. The

first subsection reviews the CPN state of art, while

the second subsection reviews the stability-aware

protocols state of art.

4.1 CPN Routing Protocol

CPN has been first introduced to create robust

routing for the wired networks (Gelenbe, 2000). It

has been tested and evaluated in later studies

(Gelenbe, 2001) to be adaptive to network changes

and congestions. A number of learning algorithms

have been researched before using Reinforcement

Learning based on Random Neural Networks

(Gelenbe and Seref, 2001). Recently Genetic

Algorithms (GA) have been used in CPN and tested

to modify and enhance paths already discovered by

SP’s to give new paths (Gelenbe and Liu, 2006).

However, studies show that it improves performance

under light traffic only and delays decisions.

A study in (Gelenbe and Lent, 2004) investigates

the number of SP’s needed to give best performance.

It resulted that SP’s in about 10% to 20% of total

data packet rate is sufficient to achieve best

performance, and that a higher percentage does not

enhance the performance. A slightly less percentage

of SP also give good results.

The study in (Gellman, 2006) compared CPN to

OSPF in IP networks. The results show that CPN

performs as good as OSPF, and it gives best routes

through learning in a very short time frame.

There have been many studies which evaluates

the CPN performance. One research studies CPN in

the presence of network worms (Skellari, 2008). It

concluded in a better failure-aware CPN. It achieved

that by introducing a detection mechanism which

stores timestamps of the last SP and ACK that pass

through the link. If no ACK was received after a SP

has passed in some determined time, the link is

considered under failure. However, there should be

an appropriate estimate for the average delay under

normal conditions for each link to be useful.

Extensions to the original CPN then followed as

research in this field continues. One extension is Ad

hoc CPN (Gelenbe and Lent, 2004), which uses

combination of broadcast and unicast of SP’s to

search for routes. The authors introduced a routing

metric “path availability” which models the

probability to find available nodes and links on a

path. Node availability was measured by the energy

stored in the node (remaining battery lifetime). Thus

SP’s selected nodes that have the longest remaining

battery with greater probability. The QoS Goal

function was a combined function of two goals:

maximum battery lifetime and minimum path delay.

The result was good performance (short delay and

energy-efficient), but there was a high number of

lost packets which meant that the algorithm did not

adapt to network changes quickly enough.

Additionally, node availability in real systems is

determined by many factors such as process load

and work environment and not just battery lifetime.

The AHCPN had some later enhancements as in

(Lent and Zanoozi, 2005) which proposed a solution

to control both energy consumption in nodes and

mutual interference of neighbouring

communications. The paper suggests an adjusted

transmission power level when transmitting DP’s

and ACK’s to save energy and reduce interference.

While SP’s are transmitted using full power. The

result is that nodes have more energy to participate

in routing. Nodes with more energy are chosen in

paths with higher probability.

Enhancements to AHCPN continued as research

developed. A new routing metric “Path Reliability”

was presented in (Lent, 2006), characterized by

reliability of nodes and links. Node reliability is

considered to be the probability that a node will not

fail over a specific time interval which is estimated

to be the average network life time. The QoS

combined goal function includes maximum

reliability and minimum path delay. Reliability is

continuously monitored, and if it drops below a

certain threshold the source node is informed to start

a new Route Discovery before links break.

4.2 Stability Aware Protocols

Stability-Aware routing algorithms aim to find the

Stability-awareCognitivePacketNetworkRoutingProtocolforMANET

longest-lived routes. However, there are many ways

to study path stability. In Associative Based Routing

(ABR) (Toh, 2000), associativity is defined to

determine a link’s connection stability and thus path

stability over space and time.

Signal Stability based Adaptive (SSA) routing

protocol depends on signal strength and location

stability (Sridhar, 2005). However simulations show

that it doesn’t perform better than a simple shortest

path algorithm. There are also extensions to this

protocol to overcome the problems (Bakht, 2005).

There is also some significant research about link

and path duration to propose that the residual

lifetime of a link determines the expected path

duration (Han, 2006). This kind of work studies the

distribution of link lifetimes in a network. Each time

a link breaks the average lifetime of the link is

updated for future use in path duration estimation.

However, the results of the study are closely related

to the mobility model assumed. It also assumes that

all nodes have the same movement pattern. It also

depends on information gathered over a long time in

order to reasonably estimate path duration and make

routing decisions accordingly.

5 METHODOLOGY

Stability-Aware CPN Routing Algorithm for

MANETs is an extension of the original Energy-

Aware Ad hoc CPN routing algorithm shown in

section 3. It has been enhanced to adapt to the

MANET environment, where node mobility raises

routing challenges. The main contribution of this

algorithm is to introduce Association Stability

degree to identify stable routes. Thus the algorithm

provides QoS through short, long-lived routes using

Computational Intelligence to make routing

decisions. This contribution enhances both the Route

Discovery Process and Route Maintenance Process.

Thus the research protocol has a faster route

discovery process. Also, it defines a route

maintenance process which is faster adaptive to

network topology changes. This section reviews the

research routing algorithm. The main operation of

the protocol are shown through these processes:

Neighbour discovery and Route Discovery,

Knowledge Acquisition and Storage, Routing Goal

and Path Reward, and Route Maintenance.

5.1 Neighbour and Route Discovery

The research routing algorithm introduces the use of

beacons to signify node existence. Nodes that are

low on battery become passive and refrain from

sending beacons. The number of beacons collected

over time from a neighbor is the degree of

Associativity for the node. Associativity degree is

reset when the node itself moves out of range of its

neighbors. The beacon interval determines the effect

on battery lifetime.

Route discovery is triggered when a node needs

to send data to unknown destination. SPs depart

from source to find a route using RNN/RL algorithm

to select the next hop by taking a unicast decision

using collected neighbor information. Each node

stores all SP identifiers which have visited it along

with source address. When SP reaches Destination,

an ACK is sent to the source and the new route will

be stored in the route cache of source node.

Transmission of data packets starts immediately.

5.2 Knowledge Acquisition and Storage

The SP’s collect specific network information as

they move around and store it in distributed fashion

as follows. Route Caches located only at source

nodes, stores complete path for all active

destinations. Cognitive Maps (CM) exist in all type

packets to store addresses and network metrics

(Battery, Arrival time, and Associativity) of visited

nodes. ACK’s distribute this information to update

mailboxes along the path. Packets store complete

route in their CM. Mailboxes are located at every

node, they keep statistics about performance of

active paths such as average delay, degree of

associativity. This information is used by RNN to

decide on next hop. Weight Tables are located at

nodes for the RNN. Weights are updated by the RL

algorithm. Neighbor Tables are created and

maintained at every node to keep information about

neighbors and their associativity degree and

forwarding delay.

5.3 Routing Goal and Path Reward

The research algorithm goal is to establish stable,

short, long-lived routes using resources efficiently.

To accomplish this goal, network information is

gathered and stored in a distributed way as shown in

the previous subsection. This information is used by

the Reinforcement Learning algorithm to determine

the Reward and Weight update according to the

delivered Qos. The Goal function to be optimized

combines Associativity degree and Delay as shown

in section 3 equation 2. The Reward is defined as the

inverse of the Goal as shown in section 3 equation 3.

IJCCI2013-DoctoralConsortium

5.4 Route Maintenance

Maintenance of previously established routes is

achieved by sending a small fraction of SP’s after

the dump packets start communicating data. Only

active routes are maintained. The use of beacons

allows fast detection of link breakage. Such that

when a node on a active path detects that the next

hop neighbor is not up, it immediately sends a

message to the source node indicating the broken

path. The source node stops sending data using the

broken path and searches its cache for other routes to

that destination. If another path is not found, a new

route discovery process is invoked.

6 EXPECTED OUTCOME

Routing in MANETs is a very challenging field.

There is no one perfect solution which fits all

MANETs. Different network characteristics need

different routing protocols. However this research is

to show that using Computational Intelligence to

find suitable routes in MANETS is experimental

with comparable results.

The research is to also show that stability-aware

cognitive routing is resilient to link failures and fast

adaptive to network topology changes. Thus the

simulation statistics to be studied are packet loss

ratio, route setup and reconstruction time, and total

packet delay. Early detection of link breakage

should avoid packet loss and decrease route

reconstruction time. Using beacons for neighbour

discovery allows a node to know its neighbours and

any link breakage should be detected early. Route

setup time should also be decreased by using

beacons due to the collected knowledge about

neighbours at each node. the research defines

simulation scenarios to show that using a goal

function based on delay alone gives unstable paths

and thus more frequent broken links and more

reconstruction of routes. While the combined goal

function of stability and delay avoids unstable nodes

without disrupting the total packet delay.

REFERENCES

E. Gelenbe, R. Lent, 2004. Power-Aware Ad Hoc

Cognitive Networks, Elsevier.

A. Murad, B. Al-, 2007. Adding QoS Extensions to

Enhanced Associativity Based Routing Protocol for

Mobile Ad Hoc Networks (MANET), American

Journal of Applied Sciences.

G. Skellari, 2009. Maintaining QoS and Reliability in

Self-Aware Networks. PhD Thesis for Prof. E.

Gelenbe, Imperial college London.

E. Gelenbe, P. Liu, 2010. Cognitive and Self-Selective

Routing for Sensor Networks, Computational

Management Science, vol. 7.

E. Gelenbe , Z. Xu, E. Seref, 2000. Cognitive Packet

Networks, IEEE.

R. Lent, F. Zanoozi, 2005. Power Control in Ad Hoc

Cognitive Packet Networks, Texas Wireless

Symposium.

R. Lent, 2006. Smart Packet-based Selection of Reliable

Paths in Ad Hoc Networks, IEEE.

E. Gelenbe, 2008. Steps Towards Self-Aware Networks,

ACM.

E. Gelenbe, R. Lent, Z. Xu, 2000. Networks with

Cognitive Packets, IEEE Computer Society.

E. Gelenbe, R. Lent, Z. Xu , 2001. Measurement and

Performance of Cognitive Packet Networks, Journal

of Computer Networks 37.

M. Gellman, P. Liu, 2006. Random Neural Networks for

Adaptive control of Packet Networks. In Proceedings

of the 16

International Conference on the Artificial

Neural Networks, Athens, Greece, 313-320.

E. Gelenbe, M. Gellman, R. Lent, P. Liu, P. Su, 2004.

Autonomous Smart Routing for Network QoS. In

Proceedings of 1

International Conference on

Autonomic computing (ICAC), NY, USA .

E. Gelenbe, P. Liu, J. Laine, 2006. Genetic Algorithms for

Autonomic Route Discovery. In Proceedings of IEEE

Conference on distributed Intelligent Systems

Washington D.C, USA.

P. Fu, J. Li, D. Zhang, 2005. Heuristic and Distributed

QoS Route Discovery for MANET. In Proceedings of

the 5

Int. Conference on Computer and Information

Technology (CIT2005), IEEE Computer Society.

E. Gelenbe, R. Lent, A. Nunez, 2004. Self-Aware

Networks and QoS. In Proceedings of IEEE.

G. Skellari, L. Hey , E. Gelenbe, 2008. Adaptability and

Failure Resilience of the Cognitive Packet Network.

IEEE Conference on Computer Communications,

Phoenix, Arizona, USA.

C. K. Toh, V. Vassiliou, 2000. The Effects of Beaconing

on the Battery Life of Ad Hoc Mobile Computers.

C. K. Toh, E. C. Lee, N. Ramos, 2004. Next Generation

Tactical Ad Hoc Mobile Wireless Networks,

Technology Review Journal, Springer.

K. N. Sridhar, M. Chan, 2005. Stability and Hop-count

based Approach for Route Computation in MANET,

IEEE.

H., 2005. Routing Protocols for Ad Hoc Networks.

Y. Han , 2006. Distribution of Path Duration in Wireless

Ad Hoc Networks and Path Selection, PhD Thesis

submitted to faculty of Department of Electrical and

Computer Engineering of University of Maryland.

E. Gelenbe , 2011. Energy-Aware Routing in the Cognitive

Packet Network. Springer-Verlag London.

Stability-awareCognitivePacketNetworkRoutingProtocolforMANET