Network Performance Project
Benchmarking of Wireless Routers
Pedro Bustamante

Noopur Chowdhary

School of Information Science
University of Pittsburgh
Pittsburgh, United States pjb63@pitt.edu School of Information Science
University of Pittsburgh
Pittsburgh, United States noc13@pitt.edu Abstract— This paper explores a basic technique for Network
Performance Analysis: Compare alternate equipment based on real measurements. Further, this work describes the comparison between two wireless routers configured as access points. The analysis of the equipment was implemented in terms of throughput (TCP) and jitter (UDP). In this way, this work aims to show the results of the benchmarking process of two wireless routers from different vendors through the utilization of a basic performance terminal tool such as “iperf”.
Keywords—benchmarking;
wireless; routers.

network;

performance;

II. GOALS
•

First, this project seeks to execute, analyze and evaluate a Network Performance Analysis based on real measurements. •

To benchmark two wireless routers of different vendors using an active software network performance tool such as iperf.

•

To measure and evaluate the total available throughput of parallel TCP connections over tested wireless routers
(acces points) in a server/client configuration.

•

To determine and exanimate the average jitter and percentage of lost packets of parallel UDP connections over the tested wireless routers (acces points) in a server/client architecture.

iperf;

I. INTRODUCTION
There is no doubt that Wi-Fi (IEEE 802.11), the most globalized Wireless Network LAN, has become one of the most noticeable ways to connect devices in universities, homes, business, etc. Nevertheless, not many users know about the complex performance factors of a Wi-Fi network, and how all the devices are able to stay connected exchanging information with the world. Moreover, the performance factors of the key element of the network: the wireless router (access point).
In this way, in the technical field (leaving the cost aside, which is most of the time the considered factor) signal strength is often the in effect performance benchmark quality of a wireless router (i.e., performance of the network).
Nonetheless, for a deeper analysis signal strength alone could not be enough to determine the viability and usability of a network. All too often, a network’s signal strength will fall within acceptable ranges, yet have insufficient data rates for an optimal (at least usable) connectivity. Considering this fact, this project aims to analyze other characteristics of wireless routers such as Throughput (TCP) and Jitter (UDP).
Indeed, this experiment setup and analysis presents the results of the comparison of two wireless routers of different vendors. The analysis of the devices would be centered in key characteristics of the most common transport layer protocols used nowadays in the internet: Transport Control Protocol, and
User Datagram Protocol.

III. GENERAL INFORMATION
A. Wireless Routers
As observed the present work is based in the benchmarking process of two wireless routers. Thus, Table I details the characteristics of the equipment to be compared.
TABLE I.
Routers
Router 1
(R1)
Router 2
(R2)

WIRELESS ROUTERS

Characteristics
Brand

Model

Netgear

WNDR4300 802.11N Dual Band

Linksys

WRT600N 802.11N Dual Band

The compared routers are commercial residential devices. In addition, they posses similar technical features (processor speed and RAM) to guarantee an objective evaluation. Further, the chosen routers are designed for medium to large residential spaces and to operate under the 802.11n standard. Besides, both devices are capable of reaching speeds between 300 to 450
Mbps (Information gathered from the vendor webpage).

Besides, both routers use the latest available property firmware, and were configured as access points to serve as a link in a client/server architecture. Consequently, the chosen wireless routers are configured undre normal parameters (no added functionalities or firmware modifications) to obtain the best possible performance under normal enviromental conditions.

Factors
TCP
Connections
UDP
Connections

Description

2

2 parallel streams / 5 parallel streams

2

2 parallel streams / 5 parallel streams

IV. METHODOLOGY

B. Used Tools
Due to the nature of this project, a tool with not only traffic generation but also measurement capability was needed.
Therefore, the natural choice was: Iperf. This performance tool is a widely-utilized open-source active software performance tool written in C. Besides, Iperf 3 (latest version) runs on several operating systems using the command window (or terminal) to test and/or evaluate a Network. It allows the creation of TCP (Transmission Control Protocol) and UDP
(User Datagram Protocol) data streams between two hosts in a
Client/Server architecture. The basic functionality of the tool is to measure the throughput (Bits/unit of time), jitter (rate of change in the end-to-end delay) and number of lost packets of the tested network (percentage).
It is necessary to point out that all the set of results were saved in text files (.txt) to keep record of the complete measurements and to facilitate the post-processing tasks.
Indeed, all the original files were processed and analyzed using
MATLAB ® and Microsoft Excel ®.
C. Factors and Variables
The first step in the experiment design is to choose the adequate response variables. In this way, this project will utilize two main response variables to analyze the performance of each of the wireless routers:
•
•

Factors and Levels for the Experiment
Levels

A. Experiment Design
As shown in Table II, 5 experimental factors with 2 levels each were chosen for the benchmarking task of this project. For each combination of factor and level, the experiment recollected 100 values of the response variables (replications).
Furthermore, the Total Number of Experiments (TNR) was obtained using the Full Factorial Design (all combinations of levels and factors); which resulted in a total of 3200 recollected values in the experiment divided in 32 sets of measurements. = ∗

)
(*+

(

= 10 22222 = 3200

(1)
(2)

B. Data collection
As explored in the previous sections, the testing was aimed under a Full Factorial Design. In this way, all the factors and its levels were combined. For instance, to obtain a determine throughput we could have: Router: 1, Frequency: 2.4, TCP, 2 parallel connections and a distance of 150ft in a set of measurements. For each of the tests (32 in total), the routers were connected as access points, serving as the link between the iperf Client (traffic generator) and the iperf Server (point of measurement) as it is shown in Fig. 1.

Throughput: Measured using parallel TCP connections
End-to-end delay variation (Jitter): Measured using parallel UDP connections.

Additionally, to guarantee a zero impact of external factors in the measurements the signal characteristics of the routers were monitored during the complete experiment.
On the other hand, the factors (parameters to be varied in the study) and its levels (values for the factors) of the experiment were also selected (Table II).
TABLE II.
Factors

FACTORS AND LEVELS

Factors and Levels for the Experiment
Levels

Description

Routers

2

Netgear / Linksys

Frequencies

2

2.4 GHz / 5.0 GHz

Distance

2

20 Feet / 150 Feet

Fig. 1. Experiment Architecture

The iperf client was configured in a single machine (Apple personal computer). In the same light, the iperf server was configured in another independent PC (Apple). Both computers had similar and sufficient technical characteristics (Operating system, processor speed, hard drive and bus capacity), which minimize any impact of the end-hosts in the results. In addition, each replication measured the throughput for each parallel stream and a resulting throughput (sum of the individual

throughputs). To fulfill the designed experiment setup, iperf
(version 3) was configured as follows:

in terms of Signal Strength. Consequently, no future analysis will be done respecting this matter.

iperf -c 192.168.1.1 -P 1 -i 1 -p 5001 -C -f m -t 100 -T 1
(Client)
iperf -s -P 0 -i 1 -p 5001 -C -f m
(Server)

The UDP configuration of the project was done in the same manner as TCP: generating traffic from client to server using the aforementioned equipment. In addition, each replication measured jitter and percentage of lost packets for each parallel stream and a resulting throughput (sum of the individual throughputs). Thus, we have the UDP configuration as follows: iperf -c 192.168.1.1 -u -P 1 -i 1 -p 5001 -C -f m -b 1.0M -t 100 -T 1
(Client)
iperf -s -u -P 0 -i 1 -p 5001 -C -f m
(Server)

C. Sets of Measurements
Table III details the combinations that were tested using the above configuration of the tool (Same for distance = 100 feet and Router 2 as well).
TABLE III.

Fig. 2. Signal Analysis (SNR (Quality) and Signal Strength)

E. Benchmarking
To analyze the individual performance of each router and its posterior comparison against each other, the following procedure was implemented:
1.

COMBINATIONS OF FACTORS (SETS OF MEASUREMENTS)

Combinations of Factors

Combination

2.

Description

1

Router 1/2.4 GHz/TCP/2 Parallel Connections/20ft

2

Router 1/2.4 GHz/TCP/5 Parallel Connections/20ft

3

Router 1/5.0 GHz/TCP/2 Parallel Connections/20ft

4

Router 1/5.0 GHz/TCP/5 Parallel Connections/20ft

5

Router 1/2.4 GHz/UDP/2 Parallel Connections/20ft

6

Router 1/2.4 GHz/UDP/5 Parallel Connections/20ft

7

Router 1/5.0 GHz/UDP/2 Parallel Connections/20ft

8

Router 1/5.0 GHz/UDP/5 Parallel Connections/20ft

3.

= = ± ∗

D. Signal Characteristics
Finally, it is necessary to point out that in each of the tests
(32 in total), the signal strength of the routers was monitored all the time. For this purpose, the embedded Apple tool “Wireless
Diagnostics Performance” was utilized. The signal characteristics that were monitored were the Signal Strength in
(dBm) as well as the Signal to Noise Ratio (dB), as shown in the examples in Fig. 2.
From this monitoring process, we can conclude that for none of the 32 tests the SNR of the Signal Strength were as critical as to be considered a constrain in the measurements. In other words, both routers displayed an acceptable performance

Verify if the measured data was correlated. Further, the coefficient of correlation was calculated for each set of measurements (32 in total)
The present project used measurements of central tendency for comparing purposes. Indeed, the following values were calculated for the analysis:
a. Total average throughput under the TCP configuration (Sum of the individual streams) b. Average jitter using the UDP configuration
(Average of the individual streams)
c. Average percentage of lost packets for the
UDP tests
Using the aforementioned values, a 90% confidence interval was calculated for each of set of measurements (32). Each of the CI was obtained through the following expression:
@
A

= 1.645 = ; = . ; = #

4.

(4)
(5)
(6)

Once a CI was calculated for each of the sets of measurements, a comparison (benchmarking) between the routers was possible. Thus, when there is no overlapping in the CI, a conclusion can be made in terms of a better performance metric of one wireless router over the other

V. RESULTS

3
2.5

A. Original Information TCP

2
1.5
1
0.5
0

1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97

As mentioned in the previous section, the TCP tests were implemented using two variations: 2 and 5 parallel streams.
Moreover, for each test the effective throughput was presented as the individual measurements and the sum of the signals (Fig.
3 and Fig. 4).

Stream 1
16

Stream 2

Fig. 5. Jitter Measurements Example (R1/2.4/UDP/2/150ft)

14
12

35

10
8

30

6

25

4

20

2

15

SUM

Stream 1

Stream 2

Fig. 3. Throughput Measurements Example (R1/2.4/TCP/2/150ft)

10
5
0

1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97

1
4
7
10
13
16
19
22
25
28
31
34
37
40
43
46
49
52
55
58
61
64
67
70
73
76
79
82
85
88
91
94
97
100

0

Stream 1

40

Stream 2

Stream 3

Stream 4

Stream 5

Fig. 6. Jitter Measurements Example (R1/2.4/UDP/5/150ft)

35

Additionally, under the UDP testing scheme, iperf also measures the percentage of lost packets (lost packets/total packets) of each stream in the connection. This characteristic is shown in Fig. 7.

30
25
20
15
10

120%

5

100%
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97

0
SUM

Stream 1

Stream 2

Stream 3

Stream 4

Stream 5

80%
60%
40%

B. Original Information UDP
The UDP measurements were configured in the same manner as the TCP tests (2 and 5 parallel streams).
Nevertheless, the response variable to analyze was jitter through the individual jitter of each stream.

20%
0%

1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
101

Fig. 4. Throughput Measurements Example (R1/2.4/TCP/5/150ft)

Stream 1

Stream 2

Stream 3

Stream 4

Stream 5

Fig. 7. Packet Lost Measurements Example (R1/2.4/UDP/5/150ft)

C. Results Router 1
All the measured information was processed for each of the routers as the first step in the comparison process. In this manner, Table IV shows the average results in terms of throughput obtained after each TCP test (8 in total) of the first benchmarked router (Netgear WNDR4300).

TABLE IV.
Frequency
(MHz)
2.4
2.4
5.0
5.0
2.4
2.4
5.0
5.0

Protocol
TCP
TCP
TCP
TCP
TCP
TCP
TCP
TCP

RESULTS ROUTER 1(THROUGHPUT)
Router 1 (Netgear)
Parallel
Distance (ft)
Connections
2
150
5
150
2
150
5
150
2
20
5
20
2
20
5
20

7.0000
6.0000

Average Throughput
(Mbits/sec)
9.6736
5.7620
39.6083
36.7984
42.8203
44.6169
100.1848
103.8958

5.0000
4.0000
3.0000
2.0000
1.0000
0.0000
1
2.4 UDP 2 150 2.4 UDP 5 150 5.0 UDP 2 150 5.0 UDP 5 150

120.00

2.4 UDP 2 20

2.4 UDP 5 20

5.0 UDP 2 20

5.0 UDP 5 20

Fig. 9. Jitter Results Router 1 (Table V)

100.00
80.00
50.00%

60.00

45.00%
40.00%

40.00

35.00%
30.00%

20.00

25.00%
20.00%

0.00

15.00%

1

10.00%
5.00%

2.4 TCP 2 150 2.4 TCP 5 150 5.0 TCP 2 150 5.0 TCP 5 150
2.4 TCP 2 20

2.4 TCP 5 20

5.0 TCP 2 20

0.00%
1

5.0 TCP 5 20

2.4 UDP 2 150 2.4 UDP 5 150 5.0 UDP 2 150 5.0 UDP 5 150

Fig. 8. Throughput Results Router 1 (Table IV)

2.4 UDP 2 20

In the same light, Table V and VI expose the average jitter and percentage of lost packets. This values were obtained for the different UDP performance tests (8 in total) of the Netgear wireless router (R1).
TABLE V.

RESULTS ROUTER 1(JITTER)

Parallel
Connections

Distance (ft)

Average
Jitter
(msec)

2.4
2.4
5.0
5.0
2.4
2.4
5.0
5.0

UDP
UDP
UDP
UDP
UDP
UDP
UDP
UDP

2
5
2
5
2
5
2
5

150
150
150
150
20
20
20
20

1.3685
6.2372
0.5872
1.8602
1.4564
4.2698
0.3693
0.5581

TABLE VI.
Frequency
(MHz)
2.4
2.4
2.4
2.4
5.0
5.0
5.0
5.0

Protocol
UDP
UDP
UDP
UDP
UDP
UDP
UDP
UDP

RESULTS ROUTER 1(PACKET LOSS)
Router 1 (Netgear)
Parallel
Distance (ft)
Connections
2
150
2
150
5
150
5
150
2
20
2
20
5
20
5
20

5.0 UDP 5 20

From the aforementioned information we can describe the performance of the first wireless router as follows:
•

Protocol

5.0 UDP 2 20

Fig. 10. Packet Loss Results Router 1 (Table VI)

Router 1 (Netgear)
Frequency
(MHz)

2.4 UDP 5 20

•

In terms of throughput we can observe that at less distance
(router to server/client) a higher throughput is obtained. In the same way the 5.0 GHz frequency of the equipment displays a higher throughput rate.
As expected, in the UDP measurements, the amount of jitter and number of lost packets is directly impacted by the number of parallel streams and the distance. Thus, more streams and distance are translated in a bigger difference in the end-to-end delay (jitter) and in more loss of packets between client and server.

D. Results Router 2
Packet Loss
(%)
0.16%
47.37%
2.26%
9.02%
0.00%
0.00%
30.43%
47.55%

Table VII shows the average results in terms of throughput obtained after each TCP test (8 in total) of the second benchmarked wireless router (Linksys WRT600N).
TABLE VII.
Frequency
(MHz)
2.4
2.4
5.0
5.0
2.4
2.4
5.0
5.0

Protocol
TCP
TCP
TCP
TCP
TCP
TCP
TCP
TCP

RESULTS ROUTER 2(THROUGHPUT)
Router 2 (Linksys)
Parallel
Connections
2
5
2
5
2
5
2
5

Distance (ft)
150
150
150
150
20
20
20
20

Throughput
(Mbps/sec)
5.3072
6.3169
5.5941
5.7859
8.9916
8.9916
6.9441
7.1983

0.00

2.00

4.00

5.0 TCP 5 20

6.00

5.0 TCP 2 20

8.00

2.4 TCP 5 20

0.00%

10.00

20.00%
5.0 UDP 5 20

2.4 TCP 2 20

Fig. 11. Throughput Results Router 2 (Table VII)

Frequency
(MHz)
2.4
2.4
5.0
5.0
2.4
2.4
5.0
5.0

Protocol
UDP
UDP
UDP
UDP
UDP
UDP
UDP
UDP

TABLE IX.
Frequency
(MHz)
2.4
2.4
5.0
5.0
2.4
2.4
5.0
5.0

Protocol
UDP
UDP
UDP
UDP
UDP
UDP
UDP
UDP

2.4 UDP 5 20

100.00%

2.4 UDP 2 20

From the aforementioned data we can describe the performance of the second wireless router in the following terms: RESULTS ROUTER 2(JITTER)

In the TCP configuration we have that: we can have more throughput rate at smaller distances. In the same way the
5.0 GHz frequency presents a higher throughput than the
2.4 band in the same router.
For the UDP measurements, the amount of jitter and number of lost packets is directly impacted by the number of parallel streams and the distance. Thus, more streams and distance are translated in a greater variation in the endto-end delay (jitter)

•
Distance (ft)

Jitter (msec)

150
150
150
150
20
20
20
20

4.5386
8.7997
5.8742
15.0149
1.7606
4.0174
4.2845
5.3822

•

RESULTS ROUTER 2(PACKET LOSS)
Router 2 (Linksys)
Parallel
Connections
2
5
2
5
2
5
2
5

80.00%

Fig. 13. Packet Loss Results Router 2 (Table IX)

Furthermore, Table VIII and IX expose the average jitter and percentage of lost packets, which corresponds to Router 2 performance indicators.

Router 2 (Linksys)
Parallel
Connections
2
5
2
5
2
5
2
5

5.0 UDP 2 20

60.00%

5.0 UDP 5 150 5.0 UDP 2 150 2.4 UDP 5 150 2.4 UDP 2 150

5.0 TCP 5 150 5.0 TCP 2 150 2.4 TCP 5 150 2.4 TCP 2 150

TABLE VIII.

40.00%

VI. ANALYSIS

Distance (ft)

Packet Loss

150
150
150
150
20
20
20
20

1.91%
5.46%
67.75%
78.70%
0.45%
0.67%
40.25%
90.33%

This final section explores the core of the project: the comparison (benchmark) of both wireless routers. In this manner, from the exposed results of both routers it is possible to calculate the 90% Confidence Interval of each set of measurements (32). Further, it is possible to determine a better performance of one over the other in each individual combination (e.g., which router has better throughput in this conditions: TCP connection on the 5.0 GHz band with 2 parallel streams at a 100 feet distance).
Therefore, for the first response variable (throughput over parallel TCP streams) we have:
TABLE X.

COMPARISON THROUGHPUT (150 FEET)
Confidence
Interval

Parallel
Connections

Average
Throughput
(Mbits/sec)

Min

Max

TCP

2

9.6736

9.341

10.0057

2.4

TCP

2

5.3072

4.799

5.8154

2.4

TCP

5

5.7620

4.646

6.8784

2

2.4

TCP

5

6.3169

5.913

6.7205

1

5.0

TCP

2

39.6083

38.022

41.1941

2

5.0

TCP

2

5.5941

5.149

6.0395

1

5.0

TCP

5

36.7984

35.137

38.4598

2

5.0

TCP

5

5.7859

5.361

6.2109

Frequency
(MHz)

Protocol

1

2.4

2
1

Router

0.00

2.00

4.00

5.0 UDP 5 20

6.00

8.00

5.0 UDP 2 20

10.00
2.4 UDP 5 20

12.00

14.00

2.4 UDP 2 20

5.0 UDP 5 150 5.0 UDP 2 150 2.4 UDP 5 150 2.4 UDP 2 150

Fig. 12. Jitter Results Router 2 (Table VIII)

16.00

Comparison
Router 1 Better
Throughput
It cannot be determined Router 1 Better
Throughput
Router 1 Better
Throughput

TABLE XI.

COMPARISON THROUGHPUT (20 FEET)

Router

Frequency
(MHz)

Protocol

Parallel
Connections

Average
Throughput
(Mbits/sec)

1

2.4

TCP

2

42.8203

2

2.4

TCP

2

8.9916

8.8639

9.1193

1

2.4

TCP

5

44.6169

44.2718

44.9620

2

2.4

TCP

5

8.9916

6.0905

7.1037

1

5.0

TCP

2

100.1848

96.3289

104.0407

2

5.0

TCP

2

6.9441

6.5781

7.3101

1

5.0

TCP

5

103.8958

102.9048

104.8869

2

5.0

TCP

5

7.1983

6.8187

7.5779

Confidence Interval
Min

Max

41.9171

43.7235

Comparison

2.4 GHz, TCP, 5 parallel streams and 150 feet. In this case, the
Confidence Intervals overlap; thus it is not possible to conclude which router has a better performance.

Router 1 Better
Throughput
Router 1 Better
Throughput
Router 1 Better
Throughput

For the second tested response variable (jitter over parallel
UDP connections) we have the following:
TABLE XII.

Router 1 Better
Throughput

COMPARISON JITTER (150 FEET)

Router

34

Parallel
Connections

Jitter
(ms)

2.4
2.4
2.4

UDP
UDP
UDP

2
2
5

2
1
2
1
2

39

Protocol

1
2
1

44

Frequency
(MHz)

2.4
5.0
5.0
5.0
5.0

UDP
UDP
UDP
UDP
UDP

5
2
2
5
5

TABLE XIII.

29

Min

Max

1.3685
4.5386
6.2372

1.3328
3.8683
5.9377

1.4042
5.2088
6.5367

8.7997
0.5872
5.8742
1.8602
15.0149

8.3099
0.5003
4.8841
1.6699
14.0068

9.2895
0.6740
6.8642
2.0504
16.0229

14

Frequency
(MHz)

Protocol

Parallel
Connections

Jitter
(ms)

1
2
1
2
1
2
1
2

19

Percentage
Lost
Packets

Comparison

0.16%
1.91%
47.37%

Router 2 More
Jitter
Router 2 More
Jitter

5.46%
2.26%
67.75%
9.02%
78.70%

Router 2 More
Jitter
Router 2 More
Jitter

COMPARISON JITTER (20 FEET)

Router

24

Confidence Interval

2.4
2.4
2.4
2.4
5.0
5.0
5.0
5.0

UDP
UDP
UDP
UDP
UDP
UDP
UDP
UDP

2
2
5
5
2
2
5
5

1.4564
1.7606
4.2698
4.0174
0.3693
4.2845
0.5581
5.3822

Confidence
Interval
Min
Max
1.3152
1.5976
1.6450
1.8763
3.7944
4.7451
3.8129
4.2218
0.3619
0.3766
3.5112
5.0577
0.5507
0.5655
5.0196
5.7448

Comparison

Percentage
Lost
Packets

Router 2 More
Jitter
Router 2 More
Jitter
Router 2 More
Jitter
Router 2 More
Jitter

0.00%
0.45%
0.00%
0.67%
30.43%
40.25%
47.55%
90.33%

9

18.00
4
0

1

2

3

R1/2.4/TCP/2
R2/2.4/TCP/5
R1/5.0/TCP/5

R2/2.4TCP/2
R1/5.0/TCP/2
R2/5.0/TCP/5

4

5

R1/2.4/TCP/5
R2/5.0/TCP/2

16.00
14.00
12.00

Fig. 14. CI Comparison (Router 1 vs Router 2) (Throughput 150ft)
(Table X)

10.00
8.00
6.00

120.00

4.00
100.00

2.00
0.00
0

80.00

60.00

1
R1/2.4/UDP/2
R2/2.4/UDP/5
R1/5.0/UDP/5

2

3
R2/2.4/UDP/2
R1/5.0/UDP/2
R2/5.0/UDP/5

4
R1/2.4/UDP/5
R2/5.0/UDP/2

5

Fig. 16. CI Comparison (Router 1 vs Router 2) (Jitter 150ft)
(Table XII)

40.00

20.00

0.00
0

1
R1/2.4/TCP/2
R2/2.4/TCP/5
R1/5.0/TCP/5

2

3
R2/2.4/TCP/2
R1/5.0/TCP/2
R2/5.0/TCP/5

4
R1/2.4/TCP/5
R2/5.0/TCP/2

5

Fig. 15. CI Comparison (Router 1 vs Router 2) (Throughput 20ft)
(Table XI)

As shown in Tables X and XI in terms of throughput over the TCP connections configuration, for all set of measurements, R1 (Netgear) presents a better performance compared to R2 (Linksys). The only exception is in the case of

As shown in Tables XII and XIII in terms of variation in the end-to-end delay (jitter) of the parallel UDP connections, for all set of measurements, R1 (Netgear) presents a better performance compared to R2 (Linksys). In other words, R1 has better variation of the end-to-end delay, which is considered a better performance especially for application layer processes running real time audio or video.
In this manner, based on the gathered measurements that were previously exposed, we can grade the performance of the wireless routers. Therefore, we can conclude, with a 90% confidence interval, that R1 (Netgear) has a better performance that R2 (Linksys) in terms of throughput (TCP), jitter and packet loss (UDP).

7.00

•

6.00

4.00

3.00

2.00

1.00

0.00
0

1
R1/2.4/UDP/2
R2/2.4/UDP/5
R1/5.0/UDP/5

2

3
R2/2.4/UDP/2
R1/5.0/UDP/5
R2/5.0/UDP/5

4
R1/2.4/UDP/5
R2/5.0/UDP/2

5

Fig. 17. CI Comparison (Router 1 vs Router 2) (Jitter 20ft)

•

VII. CONCLUSIONS

•

For the throughput response variable in TCP parallel connections, for all set of measurements,
Router 1 presents a better performance compared to Router 2. The only exception is in the case of
2.4 GHz, TCP, 5 parallel streams and 150 feet. In this case, the Confidence Intervals overlap; thus it is not possible to conclude which router has a better performance. o In the same light, for the second response variable, variation in the end-to-end delay (jitter) of the parallel UDP connections, for all set of measurements, Router 1 presents a better performance compared to Router 2. In other words, R1 has less variation of the end-to-end delay. Thus, we can conclude that the tested router Netgear
WNDR4300 has a better performance than the router
Linksys WRT600N. o 5.00

•

In addition, using Confidence Intervals (90%), this project explored the differences in performance of Router 1
(Netgear) and Router 2 (Linksys). Further, we have:

The proposed network performance study was successfully executed. Further, real measurements were taken, processed and analyzed in order to the comparison between equipment of different vendors.
The selected response variables (throughput, jitter and packet loss) were processed and analyzed from the original data, which was obtained through an iperf client/server architecture. Moreover, we observed the following circumstances: o

o

For both wireless routers in terms of throughput, we can observe that at less distance higher throughput. In the same way the 5.0 GHz frequency presents a higher throughput rate than the 2.4 band.
For the UDP measurements in the two routers, the amount of jitter and number of lost packets is directly impacted by the number of parallel streams and the distance. Thus, more streams and distance are translated in a bigger difference in the end-to-end delay (jitter)

REFERENCES
[1]
[2]
[3]

[4]
[5]
[6]

[7]

[8]

R. Golani, S. Kandalan, “WiFi Traffic Analysis Project Report”, Stony
Brook University”, New York.
Luxul Technology, “Use Iperf to Measure Wi-Fi Network Data Rates:
Data Rates vs. Signal Strength”, Lindon, UT.
J. Griffit, “How to Guide: Jperf and Iperf”, Wireless LAN Professionals, retrieved from: http://wirelesslanprofessionals.com/wpcontent/uploads/2011/02/How-to-Guide-on-JPerf-and-IPerf.pdf Iperf Support, “The Network Bandwidth Measurement Tool”, obtained from: https://iperf.fr
Netgear Support, “Manuals and Technical Information”, retrieved from: http://www.netgear.com/home/ Linksys Support, “Manuals and Technical Information”, obtained from: http://www.linksys.com/us/?gclid=Cj0KEQiA4qSzBRCq1iLhZ6Vsc0BEiQA1qt-zoQNczouEqTJHJnt2WPCMmLlnQDjVjZ02PSdmwG6RgaAkxR8P8HAQ Paessler AG, “Monitoring Tools Stay on Top of Your Network Using
Paessler’s
Monitoring
Tools”,
retrieved from: https://www.paessler.com/monitoring_tools
Kolahi, Samad S., et al. "Performance monitoring of various network traffic generators." Computer Modelling and Simulation (UKSim), 2011
UkSim 13th International Conference on. IEEE, 2011.