E.O. Karpukhin – Ph. D. (Eng.), Associate Professor, Moscow Aviation Institute (National Research University); Senior Research Scientist, Design Information Technologies Center of RAS (Odintsovo, Moscow region)
K.V. Meshavkin – Student, Moscow Aviation Institute (National Research University)
This article is about comparing efficiency of application of three correcting codes in the ARQ/FEC hybrid system. Comparing is based on three parameters - the time of data assembly, data processing intensity and code speed. Goal is to find a line, after which FEC mode will be more suitable than ARQ mode.
To obtain quantitative values of these characteristics, modeling was done using software components of the ARQ/FEC system. Software model of the system consists of a sender and a receiver connected through Netem. Netem is a network emulator that allows user to change network characteristics such as the transmission time or the probability of packet loss. Developed software model can work in two modes - ARQ mode and FEC mode.
According to the results of experiment, extended parity code effectively eliminates packet loss while probability of packet loss is low, but starting from the probability of 3% even the code with the code word length of 5 packets no longer cope with packet loss, which shows limited applicability this code in network. The average data assembly time for extended parity code stays within 10 ms on all probabilities of packet loss in the network except for 0,1%, sharply increasing to 30 ms due to a abrupt increase of length of the codeword.
A large percentage of the loss using the Reed-Solomon code happens due to the significant cost of resource processing for encoding and decoding data, which means that the receiver cannot process received packets and must discard them, otherwise this will lead to a significant increase in the data assembly time. Average data assembly time for the Reed-Solomon code stays within 15−25 ms for the packet correcting ability of 1, within 31−36 ms for the packet correcting ability of 4 and within 49−51 ms for the packet correcting ability of 8.
As for the network code, results show that some variants with certain codeword lengths has a zero probability of loss on all probabilities of packet loss in the network. Average time of data assembly by the network code stays within 35 ms, which happens because network code has minimal costs of encoding and decoding operations.
Based on results of the comparison of three correction codes, following results were obtained: in all cases except for time of data as-sembly on the probability of packet loss in the network of 0,1% the Reed-Solomon code loses to the extended parity code and network code. This happens due to the fact that the maximum possible code word length for such network parameters for the Reed-Solomon code is significantly lower than for the network code and the extended parity code, which leads to lower assembly costs, even though the Reed-Solomon code has the longest build time among all three codes. Extended parity and network code are almost equivalent in terms of build time and very similar in terms of data processing intensity.
- Kruk E.A., Malichenko D.A. Raschet zaderzhki pri ispol'zovanii kodirovaniya na transportnom urovne seti peredachi danny'x // Izvestiya VUZov. Ser. Priborostroenie. 2013. T. 56. № 8. S. 45−51.
- Karpuxin E.O., Britvin N.V. Primenenie transportnogo kodirovaniya dlya umen'sheniya vremeni sborki danny'x na priemnoj storone // E'lektrosvyaz'. 2016. № 10. S. 46−50.
- Karpuxin E.O., Gasanov E'.O., Solodovnikov V.I. Sposob upravleniya dostavkoj setevy'x paketov na storone poluchatelya dlya povy'sheniya e'ffektivnosti informaczionnogo vzaimodejstviya v telekommunikaczionny'x sistemax // Sistemy' vy'sokoj dostupnosti. 2014. T. 10. № 2. S. 33−37.
- Karpuxin E.O., Karnakov V.V. Primenenie korrektiruyushhix kodov v sistemax ARQ/FEC na transportnom i prikladnom urovnyax // Trudy' mezhd. konf. «Informaczionny'e texnologii v nauke, obrazovanii i upravlenii» (IT+S&E). 2016. S. 207−211.
- Audrius Jurgelionis, Jukka-Pekka Laulajainen, Matti Hirvonen, Alf Inge Wang. An Empirical Study of NetEm Network Emulation Functionalities // The 20th International Conference on Computer Communications and Networks (ICCCN). 2011.
- Karpuxin E.O. Variant ARQ/FEC sistemy' dlya protivodejstviya blokirovkam paketov v ocheredyax na priemnoj storone // Informaczionny'e texnologii i vy'chislitel'ny'e sistemy'. 2016. № 4. S. 5−12.
- Toke Høiland-Jørgensen, Bengt Ahlgren, Per Hurtig, Anna Brunstrom. Measuring Latency Variation in the Internet // The 12th International Conference on Emerging Networking Experiments and Technologies. 2016. P. 473−480.