Now that the activities of 802.11n standard have been finalized and IEEE 802.11 wireless cards are available, we can deploy a real test-bed and retrieve performance measurements for various scenarios. In addition, the open-source community has devoted huge efforts to provide Linux based firmwares, such as carl9170 [177],
ar9170 [178] and brcm80211 (by Broadcom) [179]. Open Source firmwares are suitable for providing the easiest approach to implement an experimental method while at the same time supporting a great number of functionalities within the framework of the respective hardware platform used.
The proposed ADCA and SDCA can be extended even further. Both algo- rithms, in their own manner, are trying to speculate the presence of a TCP flow and determine its parameters, e.g. TCP Window size. For the SDCA extension, we propose a basic but still innovative flow-features based classification mecha- nism with evaluation experiments showed positive results. Additional classifying methods have been introduced, such as Support Vector Machine (SVM) algo- rithm, which achieves a 98.0% accuracy on every trace and application[153]. The basic principle of SVM is to construct the optimal separating hyperplane, which maximizes the distance between the closest sample data points in the (reduced) convex hulls for each class, in an n-dimensional feature space[180]. Therefore, DCA is a subject of extended research that could easily improve performance even further by implementing new approaches.
The initial DCA algorithm was firstly introduced as a concept in TGnSync Proposal Technical Specification Document [181]. TgnSync was one of the main industry associations that took part in the development of the IEEE 802.11n standard. The standard’s specification document had to undergo through many changes and various balloting sessions before eventually reach its final form as we know it today. Usually during a standardization process, there are multiple issues that to have to be addressed and negotiated between the Task Group’s members, mainly technical differences but also licensing of intellectual property which then could lead to endless conflicts. Nevertheless, when TGnSync and an- other consortium, known as WWiSE, collaborated together, the DCA proposal was dropped from their joint specifications document. The reasoning is unknown but we assume that DCA design was on its initial phase and hadn’t been re- searched thoroughly. Evidence of the issues that the algorithm imposes is the TCP Window size problem, also discussed in this text. However, the evolution of the HT wireless broadband networks still continues with the emerging devel- opment of IEEE 802.11ac and IEEE 802.11ad. Both end specifications will aim to enable multi-station WLAN communication at multi-gigabit speeds. Frame
aggregation will still be a key technology for this future standards and so can the inclusion of well defined DCA algorithm.
Final but not least, since energy-related considerations are gaining popularity in wireless networks, especially for mobile devices, there is a tremendous interest in energy efficiency. The main transmission technique in 802.11n is utilizing the MIMO technology, enabling the use of multiple sending and receiving antennas with the objective of providing high rates but resulting in higher power consump- tion too. Early studies have shown that enhanced frame aggregation schemes that increase channel utilization while supporting robust frame delivery, can also re- duce the energy cost for wireless devices [182]. The improvements from the DCA operation can also be compared with possible gains over the power consumption as well.
Usage Models
A general definition for the term “Usage Model” is given within IEEE’s Usage Models documentation for the emerging 802.11n amendment [68]. According to 802.11 TGn, usage model is a specification of one or more applications and environments from which a simulation scenario can be created once the traffic patterns of the applications are known. A use case is a description of how end users uses an application, such as HDTV, video streaming, internet transfer, VoIP and etc. and how these users are deployed over the system. In general, usage models are created to cover various market-based use-cases and intend to support the definitions of network simulations that will allow 802.11 TGn to evaluate performance of various proposals in terms of network throughput and goodput, delay, packet loss and other metrics.
The following usage models are enumerated according to [68] and brief de- scriptions and definitions are provided.
A.1
Model 1 - Residential
The first scenario represents an indoor (room to room) residential network with several HT devices. Wireless connectivity has been spread over a residential platform for a long time now, a distinguished example is the use of cordless telephones that can provide the flexibility to move around the house and have conversations on the phone with minimum jitter. Nowadays, more and more
home wireless devices are being used and further more are being developed for the near future. By introducing higher data rates and QoS, users will be able to view SDTV and HDTV anywhere in the house and simultaneous talk on their VoIP telephones, surf on the Internet, listening to MP3 music that is stored on a central wireless unit or even playing games on-line via their wireless consoles.
Figure A.1: Spatial distribution in OPNET for Usage Model 1
STA Name Role Dest. STA Mean Rate Rate Distrib. MSDU Delay Application AP Access Point STA 1 19.2 Mbps Constant, UDP 1,500 B 200 ms HDTV
STA 3 24 Mbps Constant, UDP 1,500 B 200 ms HDTV STA 4 4 Mbps Constant, UDP 1,500 B 200 ms SDTV
STA 4 1 Mbps TCP 300 B Internet File
STA 7 0.096 Mbps Constant, UDP 120 B 30 ms VoIP STA 8 0.096 Mbps Constant, UDP 120 B 30 ms VoIP STA 9 0.096 Mbps Constant, UDP 120 B 30 ms VoIP
STA 10 2 Mbps UDP 512 B 200 ms Internet
Streaming
STA 11 0.128 Mbps UDP 418 B 200 ms MP3 Audio
STA 1 HDTV Dis-
play
AP 60 kbps Constant, UDP 64 B 100 ms VoD Control Channel
STA 3 HDTV Dis-
play
AP 60 kbps Constant, UDP 64 B 100 ms VoD Control Channel
STA 4 SDTV Dis-
play, Gaming & Printing
STA 10 30 Mbps Constant, TCP 1,500 B Local File Transfer
STA 5 Video Phone STA 6 0.5 Mbps Constant, UDP 512 B 100 ms Video STA 6 Video Phone
& Internet Upload
STA 5 0.5 Mbps Constant, UDP 512 B 100 ms Video
STA 7 VoIP Phone AP 0.096 Mbps Constant, UDP 120 B 30 ms VoIP STA 8 VoIP Phone AP 0.096 Mbps Constant, UDP 120 B 30 ms VoIP STA 9 VoIP Phone AP 0.096 Mbps Constant, UDP 120 B 30 ms VoIP STA 10 Video Con-
sole & Internet En- tertainment
AP 1 Mbps Constant, UDP 512 B 50 ms Console to
Internet
STA 11 Video Gam- ing Con- troller
STA 10 0.5 Mbps Constant, UDP 50 B 16 ms Controller to Console
The main role that each device possesses during this scenario can be found in Table A.1. Some of these stations may operate more than one application depending on their functionalities. The complexity of the scenario’s configuration increases while we consider direct links between STAs and with the AP who also acts as a flow coordinator. The spatial distribution for the stations over a residential plot of 20m range is shown in Figure A.1.