Increased throughput of heavily loaded WiFi networks by 4-7 times
Scientists from the University of North Carolina have developed and tested an algorithm that allows, according to experiments, to increase the capacity of wireless networks under critical load several times, where many people use them at the same time - in cafes, airports, conference rooms. In such conditions, the number of lost packets sharply increases, which quickly leads to an avalanche reduction in throughput (sometimes tenfold).
A modification of the firmware of the access point, called WiFox , allows to reduce the number of collisions under high loads, thereby avoiding congestion in an overloaded network. A detailed description of the algorithm will be presented at the conference on experimental network technologies CoNEXT , which will be held in France from 10 to 13 December.
Based on a brief description of the principle of WiFox, we can assume that we are talking about the modification of the Quality of Service algorithms, which provides a more uniform distribution of bandwidth between network participants and the priority use of a radio channel by an access point if a jam occurs in its output buffer. The authors of the system compare WiFox with a policeman on a heavily loaded junction.
Scientists have tested the algorithm in the laboratory, in networks with 25 and 45 devices connected to one access point. WiFox shows the best results, the higher the load on the network: for 25 users, the gain was 400%, for 45 - 700%. In addition, the delay time decreased by 30% -40%.
')
The problem of bandwidth of wireless networks today is becoming more acute - the number of devices with WiFi is growing very fast, the requirements for speed and the absence of delays increase. Changing protocols and standards at the physical level is very expensive and is implemented slowly, so software optimization methods are very much in demand. So, recently, a group of researchers from the Massachusetts Institute of Technology has demonstrated the use of an error correction algorithm on TCP networks, which solves the same problem as WiFox, but in a different way - instead of avoiding packet loss, it allows you to recover much of the information lost. redundancy, thereby preventing an avalanche of repeated packet forwarding.
UPD: Habrayuzer apcsb found the link to PDF with this work.
A modification of the firmware of the access point, called WiFox , allows to reduce the number of collisions under high loads, thereby avoiding congestion in an overloaded network. A detailed description of the algorithm will be presented at the conference on experimental network technologies CoNEXT , which will be held in France from 10 to 13 December.
Based on a brief description of the principle of WiFox, we can assume that we are talking about the modification of the Quality of Service algorithms, which provides a more uniform distribution of bandwidth between network participants and the priority use of a radio channel by an access point if a jam occurs in its output buffer. The authors of the system compare WiFox with a policeman on a heavily loaded junction.
Scientists have tested the algorithm in the laboratory, in networks with 25 and 45 devices connected to one access point. WiFox shows the best results, the higher the load on the network: for 25 users, the gain was 400%, for 45 - 700%. In addition, the delay time decreased by 30% -40%.
')
The problem of bandwidth of wireless networks today is becoming more acute - the number of devices with WiFi is growing very fast, the requirements for speed and the absence of delays increase. Changing protocols and standards at the physical level is very expensive and is implemented slowly, so software optimization methods are very much in demand. So, recently, a group of researchers from the Massachusetts Institute of Technology has demonstrated the use of an error correction algorithm on TCP networks, which solves the same problem as WiFox, but in a different way - instead of avoiding packet loss, it allows you to recover much of the information lost. redundancy, thereby preventing an avalanche of repeated packet forwarding.
UPD: Habrayuzer apcsb found the link to PDF with this work.
Source: https://habr.com/ru/post/158949/
All Articles