Fuzzy RED: Congestion control for TCP/IP Diff-Serv

Fuzzy RED: Congestion control
for TCP/IP Diff
-
Serv

•
Explosion of Internet
–

new congestion control
method is needed

•
WHY? :

–
Users now demand for integrated services network

–
New services with high bandwidth demands (video on
demand, video conferencing, etc)

–
Need for quality of service demanded by new
applications (e.g. streaming video)

–
Current internet infrastructure can only support best
effort traffic

Diff
-
Serv

•
A more evolutionary approach than Int
-
Serv

–
Int
-
Serv had connection establish overheads.

–
In Internet based applications they were in some cases
bigger than the existing connection period

•
Diff
-
Serv does not require significant changes to
the Internet Infrastructure

•
Uses existing ToS bits in IP header of service
differentiation

•
Works in the edges of a network

•
Provides QoS using drop
-
preference algorithm

Diff
-
Serv

•
Differentiation of services is provided
through 3 classes of services called per
-
hop
behaviour

•
Expedited Forwarding
–

EF:

–
Low losses

–
Very low queuing delays

–
Allocation of resources through SLA at
connection setup

Diff
-
Serv

•
Assured Forwarding
–

AF:

–
Low packet losses

–
Has 3
-
4 independent forward classes

–
Each such class has 2
-
3 different drop
preferences

–
Preferentially drops best
-
effort packets and
non
-
conforming packets when congestion
occurs

RED

Most popular algorithm used for Diff
-
Serv networks:



for each packet arrival


calculate the average
queue size avg

if min
th



avg



max
th


calculate probability
p
a

with probability
p
a
: mark the arriving packet

else if
max
th




avg


mark the arriving packet

Explain what mark means


RED

•
Uses min, max dropping thresholds for each
class

•
The algorithm used for calculating the
queue average determines the allowed
degree of burstiness

•
The p
a

probability is a function of the
average queue size. Varies linearly from 0
to 1.

RED

EF

AF Class

Best Effort

Check and

traffic

shaping

Priority Queue

Yes

No

Discard

RIO Queue

Min

Max

Fuzzy RED

•
We replaced fixed thresholds with an FLC

•
FLC
-

Fuzzy Logic Controller:

–
Calculates p
a

based on two inputs, queue size
and queue rate of change

•
The two inputs are described by fuzzy sets

•
The FLC determines the p
a

by applying a
set of rules.

•
Each class of service has an FLC


Fuzzy RED
-

The algorithm

Algorithm:

for each packet arrival


calculate queue size, queue rate of change

calculate probability p
a
based on above metrics

with probability p
a

mark the arriving packet


p
a

is calculated by the FLC


Fuzzy RED
–

Input Sets

•
FLC set for queue
-

q (90 packet buffer)

empty 0 0 18 35

moderate 20 33 42 63

full 44 64 90 90

•
FLC set for queue rate of change
-

dq

decreasing
-
44
-
44
-
7
–
1

zero
-
14 0 0 14

increasing 1 7 44 44

Fuzzy RED
–

Output Set

•
FLC set for p
a


zero
–
20
–
20 0 0

low 0 0.10 0.10 0.20

medium 0.15 0.20 0.20 0.30

high 0.20 0.60 0.60 1.0

•
As with input sets p
a

can also be different for each
class of service

•
Best describe the behavior expected by the
network administrator

Fuzzy RED
-

Rules

•
Based on linguistic rules we calculate the
dropping probability

•
Each class of service has its own definition
of set and rules

•
Sample of rules:

–
If q is empty the p
a

is zero

–
If q is full and dq is zero the p
a

is medium

–
If q is full and dq is increasing then p
a

is high

Fuzzy RED
–

Calculating p
a

Input Evaluation

Output Evaluation

Rule: If q is full and dq is decreasing then

pa is high

44 64 90
-
20
-
7
–
1 0.2 0.6 1


Input: q is 68 and dq is
-
3

Output: the weighted average of the colored surface: 0.4

1

0.4

Fuzzy RED
-

Conclusions

Advantages:

•
Simplicity
-

just 3 steps

•
Effectiveness

•
Scalability
-

each class has its own FLC rule
file

•
Robustness

•
Currently ………