You are here
WiFi throughput research during the wireless video transmission
Based on the IEEE 802.11 network equipment there might be designed comparatively cheap systems for video transmissions by wireless. However, if any video sources are available in a such system, there might be the problems which cause an inefficient utilization of the channel. It is complicated to determine what is the actual number of the sources of the video can be used in the system using only the parameters of a manufacturer of the network equipment and video codecs properties. The purpose of this research is to evaluate efficiency of the channel utilization and to propose the methods how to improve it.
We consider the system of the multiple access to shared wireless communication channel , which includes a large number of similar type users (digital video cameras) and the receiving/processing center. The connection of an abonent with the center is supported by the wireless channel based on the IEEE 802.11 standard. In this case the users transmit the precompressed data via the WiFi adapters and receiving/processing center receives this data through access point. We make the following assumptions:
The capacity (throughput) of the channel is sufficient for the overall stream transmission from M abonents provided that we do not include the overhead caused by the multiple access problems (channel utilization by several abonents).
The video stream from each abonent camera can be viewed as a sequence of the comparatively short inter frames which are mixed with the longer key (intra) frames (for video transmission the MPEG2, MPEG4 and H.264/AVC are used)
Estimation of actual system throughput.
The basic mode which is supported by all the 802.11 standard equipment is Distributed Coordination Function (DCF) mode. This mode is based on the random multiple access to the channel. In a DCF access some conflicts may arise. They cause the situation when the actual number of the users in the system should be reduced to fulfill the mentioned in assumption. Some new methods should be worked out in order to estimate the allowed number of users in this system with conflicts. These methods should take into account:
the set of requirements to the video quality (QoS) received by receiving/processing center;
statistical properties of the compressed video stream according to the characteristics of the used bit-rate control algorithm;
characteristics of the collision detection algorithm implemented in IEEE 802.11 equipment.
Introduction of methods to improve the actual throughput of the system.
The goal of this research is to achieve either the best possible quality of the transmitting video or the maximum number of the users in the system for a limited channel resources to upgrade the basic system.
If case of no motion the camera transmits rather small inter frames. During such transmission the collisions are seldom. If there is motion then the size of frames increase. If motion is intensive then the key frames is formed. The size of a key frame is 4..8 times bigger than the size of an inter frame. If several cameras form the key frame simultaneously and transmit it, then in most cases the collision occurs. This collision will be resolved for a long time.
There are three ways to avoid such collision and to improve the system throughput. The first way is to use desynchronization scheme for transmitting the key frames to several users.
The second way is to redistribute the channel resources considering the users with intensive motion in the camera input by using video compression rate control algorithm.
The third way is to avoid using the random access mode and to use the mode with access point coordinating the users network - Point Coordination Function (PCF) mode. In this mode collisions are completely eliminated, but a part of the channel resources will be spent for transmitting the control frames from access point to the users.
In the process of working out the project the theoretical research should be executed, the demo stand should be created in order to demonstrate the proposed approach. It should be taken into consideration that most network adapters and access points do not support the Point Coordination Function mode. The corresponding equipment should be provided to create such a stand for demonstration purposes.