A Practical Experiment with Draft-802.11n and 802.11g


Download PDF

A Practical Experiment with Draft-802.11n and 802.11g [PDF 159KB]


Introduction

How good is the new 802.11 draft-n protocol in comparison to the older, more familiar 802.11g? What value do you receive when you upgrade your present access? The experiment detailed in this paper sets out to determine the answers to these questions. High definition video clips were timed as they streamed over a wireless connection to test for smoothness and playability. Included is a discussion about QoS and WMM, processes that prioritize data packets based on content to allow video streams and Voice over Internet Protocol (VoIP) packets higher priority than normal data transfers.


The Experiment

The purpose of this experiment was to determine if video streaming with 802.11N can gain the average computer owner smoother, more effortless transmissions than streaming with 802.11G. A controlled environment was used to prevent unintended interference. This environment consisted of a small wireless network not attached to the internet. A server was hard wired to the router to eliminate delay times from server to router. Two routers were employed; an 802.11g router was used first and then a draft-802.11n router. Wirelessly, an Intel® Centrino® 2 with vPro™ technology-based computer was connected to the network using a draft-802.11n adapter card.

The process was to transmit HD video streams over each router. These streams included some groups of the same video stream at different bit-rates and resolutions. An actual time for each stream was recorded. These times included any buffering or delays in transmissions. Observations during video transmission were noted. Since virtually all videos required an initial buffering time, those seconds were recorded visually. A second test was conducted to check the quality of the video streaming when other data was downloaded during video transmission.


The Setup

This testing was done in an ordinary house and designed to highlight what the ordinary person could expect to achieve.

Hardware: The equipment used was as follows:

  • Server with Windows Media Center* operating system loaded with HD video streams at various bit-rates and resolutions.
  • Intel Centrino 2 with vPro technology Software Development Platform (SDP) which includes an 802.11 draft-N adapter (Intel® Wireless WiFi Link 4965AGN) with Windows Vista* operating system.
  • Linksys Wireless-G Broadband Router*, Model # WRT54G v5
  • Linksys Wireless-N Broadband Router*, Model # WRT300N

 

Both the adapter and the draft-N router were versions 1.0.

Software: The software included Microsoft Media Player that played the videos on the Intel Centrino 2 with vPro technology-based computer and a small freeware timer application to allow play times to be captured. A large movie file that was 2.12 GB in size was used as the downloading file.

The server and routers were set up in a separate room approximately 50 feet from the Intel Centrino 2 with vPro technology-based computer. To prevent any server-to-router delay in data transmission, the server was hard wired to the routers by means of a network cable. Initially, the routers were connected to the internet in order to configure them, but later, during streaming, the internet connection was removed to prevent possible interference.

The high definition videos used are listed in Table 1:

Table 1. List of High Definition Videos Used for Experiment

Video Title

Specification

Video Length

Alexander Trailer

720p

1:53

Amazon

720p

1:42

Coral Reef Adventure

720p

1:45

Coral Reef Adventure

1080p

1:45

Digital Life WMP10

720p

1:03

Discoverers

720p

2:31

Dolphins

720p

1:46

Experience WMP10

720p

1:00

Journey Into Amazing Caves

720p

1:27

Journey Into Amazing Caves

1080p

1:25

London HD 1:3 MBPS

1.3 MBPS

3:12

London HD 10 MBPS

10 MBPS

3:04

London HD 20 MBPS

20 MBPS

3:04

The Living Sea

720p

2:49

The Living Sea

1080p

2:47

 

The three “London” videos are especially interesting as they were encoded at various speeds indicated by the notation under the “specifications” column. The video encoded at the 1.3 Mbps speed was indistinguishable from the other 720p and 1080p videos in the list. The 10 and 20 Mbps videos, however, showed jagged edging around parts of the screen where fast movement was happening, such as the motorcycle going by quickly, as shown in Figure 1.

Figure 1. Jagged Edges Around Fast-Moving Objects

Apparently, this type of blurring is due in part by the technique used to compress the frames together at a particular frame rate. A high-end graphics card might eliminate this phenomenon completely.


Collecting the Data

The data collected during this experiment was divided into four sections:

  • Wireless-G Broadband Router with streaming video alone.
  • Wireless-G Broadband Router with streaming video and 2.12 GB file download.
  • Wireless-N Broadband Router with streaming video alone.
  • Wireless-N Broadband Router with streaming video and 2.12 GB file download.

 

Virtually every one of the video streams required some few seconds of buffering time during the start of play no matter which of the two routers were in place. The beginning buffer time was recorded for informational purposes when analyzing the amount of drag, stop, jumps, or buffer time found during the actual playing of the video. Please make note that this number was caught on the fly and is subject to human error. Indeed, the final time is also subject to human reaction time and this should be taken into account while examini ng the results.


The Results

What follows is the distillation from the actual data that can be found in Appendix A. The improvement of video streaming times when enjoying the full bandwidth of the router was only an average of 4% better with the draft-802.11n router than with the 802.11g router. The draft -802.11n router and network adapter showed their better performance when video streams played while simultaneously downloading a large file. The 802.11n-type router overall performed 4X faster than the 802.11g protocol router when sharing bandwidth with large file downloads.

The figures above don’t really describe the whole story. The two higher than normal Mbps videos, “London HD 10 MBPS” and “London HD 20 MBPS” respectively, showed the most impressive results, obtaining 10X and 4X speed improvement individually. Once these figures are subtracted from the overall result, the improvement for the 802.11n protocol devices falls to 2X. Still, a good performance increase for a new protocol that is not yet ratified. It is estimated that ratification of the 802.11n standard will occur somewhere around June of 2009.

The data shown below in Table 2 is a wrap-up of the Appendix A data. The averages of the three runs for each router are listed and the background of whichever of the two data points is smaller was highlighted in pale green. Bright yellow indicates performance improvements of 2 times or greater. Looking at this table it is plain to see that if you are planning to stream video while you download other files then the Draft-N router and adapter would be a very good choice.

Table 2. Average Drag Time Extracted from the Data in Appendix A

Video Name

Video Streaming Alone

Streaming with File Download

Average G

Average N

X Improve-ment

Average G

Average N

X Improve-ment

Alexander Trailer_720p

0.513

0.575

0.892

41.293

4.451

9.277

Amazon_720p

0.333

0.207

1.607

4.533

3.797

1.194

Coral Reef Adventure_720p

0.625

0.526

1.187

29.290

10.082

2.905

Coral Reef Adventure_1080p

0.846

1.372

0.616

2.977

1.508

1.974

Digital Life WMP10_720p

0.473

0.407

1.162

2.377

2.377

1.000

Discoverers_720p

1.435

1.195

1.201

5.261

4.098

1.284

Dolphins_720p

0.310

0.523

0.592

14.609

5.281

2.766

Experience WMP10_720p

0.783

0.448

1.746

1.872

1.985

0.943

Journey Into Amazing Caves_720p

0.159

0.434

0.366

3.465

3.519

0.985

Journey Into Amazing Caves_1080p

1.119

0.427

2.618

4.227

2.429

1.740

London HD 1:3 MBPS

0.919

-0.214

-4.294

2.188

1.338

1.635

London HD 10 MBPS

0.746

1.750

0.426

129.103

11.998

10.761

London HD 20 MBPS

49.854

48.586

1.026

384.794

94.543

4.070

The Living Sea_720p

0.340

3.527

5.410

6.637

0.815

The Living Sea_1080p

1.451

1.712

0.847

6.713

7.935

0.846

Averages

4.051

3.886

1.042

42.541

10.799

3.939

 


Quality of Service and Wi-Fi Multimedia

What does Quality of Service (QoS) mean? To the uninitiated, QoS might bring to mind a pair of rabbit ear antennas being twisted this way and that in order to produce the best signal. When discussing wireless QoS, however, the meaning may be less clear. The transfer speed of media to and from a device determines how quickly data is sent from one machine to another. The access point puts all of the data packets going to a certain device into a queue using a First In First Out method. That means that all the data from one source may be scattered between other packets waiting for the same chance to be sent to the destination device. Without some plan or method of deciding what packets are more critical to the user, all packets have the same importance in the eyes of the access point.

For many years, traditional wired Ethernet LAN has had a packet priority assignment and management system, which prioritizes packets being sent on the network. Implementing this system is known as establishing Quality of Service (QoS) by increasing the speed that higher priority packets are pushed through the network. The protocol 802.1p established eight different levels of priority, including a default priority “0” that is called “best effort.”[1] This number is determined by the media access control (MAC) at layer 2 of the OSI networking model. The assigned level, however, is lost as soon as the packet leaves the LAN.

When connected to the internet by wire, the Ethernet cable is one access point (AP). If only one device is connected to the internet by that cable then it is using the entire bandwidth of that AP. A wireless access point is somewhat different, as all devices that connect through that, point must share the available resources. As more devices connect, the amount of bandwidth per device shrinks. Without a priority system, all packets have equal status and a first-come, first-served system takes over. VoIP, high-end gaming, and media applications cannot transmit smoothly unless their packets are delivered in a consistent and timely manner. In order to ensure this, another method was needed to prioritize the packets for wireless communication.

Until the newly certified 802.11e standard is in wide use, WMM*, certified by the WiFi Alliance* has been filling the gap. WMM is defined by the WiFi Alliance as a profile or subset of 802.11e. What WMM does is to set a contention window that is based on the service type of the traffic and the amount of wait time for MAC layer medium access. There are four different contention levels; “voice”, “video”, “best effort” and “background”. Assignments of levels are set for voice, audio and video packets only. The “voice” category has the lowest wait time and therefore highest priority.[2] With WiFi Multimedia, VoIP, videos and music can work more smoothly on wireless connections, reducing delays, jitter, and pauses while providing a superior user experience.[3]

The 802.11e standard employs the same basic concept as WMM but refines the contention levels into more Traffic Classes (TC). Channel access is controlled by one of two methods; Enhanced Distributed Channel Access (EDCA) or Hybrid Coordinator Function (HCF) Controlled Channel Access (HCCA). With EDCA, each traffic class has an assigned Transmit Opportunity (TXOP) that allows it to send as many frames as can fit into the maximum allowable TXOP period. A large frame, too big for a single TXOP period, is separated into smaller sections. HCCA improves upon the idea of Contention Free Periods (CFPs) and Contention Periods (CPs) presented in the first 802.11 MAC. With the new protocol, a CP can be interrupted for a CFP most of the time. When a Contention Period is in force, EDCA is used. Otherwise, the access point coordinates the sending of packets. In addition to Traffic Classes, Traffic Stations (TS) are also described in HCCA. If more than one station is polling the access station, HCCA can collect information on what station has the most high priority packets waiting and therefore deserves access time soonest.[4] Both of these MAC layer protocols improve the Quality of Service found with the draft-N 802.11 technology.


Conclusion

The purpose of the experiment described in the previous pages was to determine if the average person, setting up a wireless video streaming connection in his or her own home, would be able to see the benefits offered with the draft-802.11n technology. Although not conclusive by any means, it appears that the value of the new standard is seen most clearly, when bandwidth is being shared between two or more tasks. The improved Quality of Service provisions allow video streaming and Voice over IP calls priority in send and receive queues and this, in and of itself, may be reason enough to make the move to 802.11n.


About the Author

Judy M. Hartley is a Software Application Engineer in the Software Solutions Group. She earned her B.S. in Computer Engineering at Virginia Tech in 2001. Right after graduation, she began working for Intel, moving almost 3000 miles to Chandler, Arizona. As a Product Development Engineer, she worked on testing and production readiness of set-top box chips. Because she missed the rain, in 2005 she relocated to one of Intel’s northwest locations in order to turn her attention to software and software enabling.

 

Appendix A: Data Sheets

 

[1] Quality of service, Wikipedia entry, http://en.wikipedia.org/wiki/Quality_of_Service.

[2] Barkay, Omri and Ben-Shalom, Omer; Implementing Quality of Service for Voice over Wireless LANs, August 2006, /sites/default/files/m/e/b/e/implementing_qos_for_voice_over_wireless_lans.pdf.

[3] VoWLAN – Voice over Wireless LAN, Intelpedia

[4] IEEE 802.11e-2005, Wikipedia entry, http://en.wikipedia.org/wiki/IEEE_802.11e-2005.

 

Categorias:
Para obter mais informações sobre otimizações de compiladores, consulte Aviso sobre otimizações.