IEEE Consumer Electronics Magazine - July 2018 - 62

in the Floodlight and Rosemary
controllers, the bandwidth started
decreasing rapidly as the attack
rate packets per second increased
in both environments.
Distarch-sCnet netwOrk
PerFOrmanCe evaLuatiOn

DistArch-SCNet MODEL EFFiCiENCy

DistArch-SCNet MODEL DEFENSE EFFECTS
To evaluate the defense effects of each SDN controller during
saturation attacks in the proposed distributed blockchain network model, we performed our experiments in hardware and
software environments. We used the LinkSys WRT54GL
commercial OpenFlow switch for the hardware environment
and Mininet for the software environment. We also used a
POX controller as a software platform to operate the controller. To improve performance, we set up two different queues
for communication from the switch to the controller and from
the controller to the switch. To send a floating user datagram
protocol attack to the switch, we configured a client as an
attacker and let other clients communicate normally. For the
evaluation of the impact on bandwidth at different attack rates,
we ran our proposed model and compared it with the Floodlight and Rosemary controllers in both environments.
To minimize any unintended overhead, we ran each network controller at a time. Figure 4(a) and (b) shows the bandwidth at different attack rates in the hardware and software
environments. In the Floodlight and Rosemary controllers, the
bandwidth started decreasing rapidly as the attack rate packets

In this section, we evaluate the efficiency of the proposed DistArch-SCNet network model using the throughput, response
time, and incurred-delay performance metrics. We gathered
real-time traces of our own distributed blockchain-based data
center and the Amazon EC2 cloud data center. We configured
our distributed data center with eight desktops, each of which
had 64 GB of DDR3 RAM, and an Intel i7 processor in our
research lab. To verify the effectiveness of our proposed
ECHA algorithm, we captured the experimental data of our
proposed DistArch-SCNet model with and without the ECHA
algorithm and compared it with the core model of the data
center. Figure 5(a) demonstrates the delays incurred with and
without the ECHA algorithm in the proposed DistArch-SCNet
model and compares it with the core data center infrastructure. This figure shows that the response time of the proposed
DistArch-SCNet model with the ECHA algorithm is reduced
compared to the DistArch-SCNet model without the ECHA
algorithm and that it has improved and performed linearly
compared to the core data center infrastructure.
The variation in the delay associated with the number of
requests generated by the installed clients is illustrated in Figure 5(b), which demonstrates an increase in the delay when the

1.8

1.6
1.4

Bandwidth (Gb/s)

per second increased in both environments. As the attack rate
increased, the OpenFlow switch started to malfunction because
of the buffer overflow and lack of proper methodology to handle the attack by the controller and update the flow table
accordingly. On the other hand, our proposed model effectively
handled the saturation attack and showed superior performance
compared to other methods in terms of bandwidth. In our proposed model, the bandwidth decreases progressively in the
hardware environment due to the lack of addressable memory
of the ternary content in our switch.

7
6
5
4

1
0.8
0.6
0.4

3
2

1.2

0.2
0

200

400

600

800

1,000

1,200

Attack Rate (PPS)

200
300
400
Attack Rate (PPS)

(a)

(b)
DistArch-SCNet

Floodlight

100

500

600

Rosemary

figurE 4. A bandwidth with and without using the proposed model during saturation attacks in the (a) hardware environment
and (b) software environment.

62 IEEE Consumer Electronics Magazine

july 2018

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - July 2018