Convolutional Codes
Author
Bregnballe, Esben Dissing
Term
4. term
Education
Publication year
2024
Submitted on
2024-12-18
Pages
23
Abstract
Når data sendes over upålidelige forbindelser, kan der opstå fejl. Fejlrettende koder tilføjer ekstra, redundant information, så modtageren kan opdage og rette fejl. En klassisk metode, lineære blok-koder, opdeler data i faste beskeder og tilføjer redundans; hver besked plus redundans udgør en kodet blok (codeword). For at afkode korrekt skal man vide præcist, hvor én kodet blok slutter, og den næste begynder. På kanaler, der kan tabe bits uden varsel, kan denne justering gå tabt. Konvolutionelle koder er udviklet til at håndtere netop dette. I stedet for adskilte blokke koder de data som en kontinuerlig strøm. Hvert udsendt symbol beregnes ud fra den aktuelle indgang og et antal foregående symboler. Ved at fordele information over nabosymboler søger konvolutionelle koder at bevare sammenhæng i strømmen og håndtere tab af blokgrænser på kanaler, der kan slette bits.
When data travels over unreliable connections, errors can occur. Error-correcting codes add extra, redundant information so the receiver can detect and fix mistakes. A classic approach, linear block codes, splits the data into fixed-size messages and appends redundancy; each message plus redundancy forms a codeword. Decoding these codes requires knowing exactly where one codeword ends and the next begins. On channels that can silently drop bits, this alignment can be lost. Convolutional codes are designed to address this situation. Instead of separate blocks, they encode the data as a continuous stream. Each output symbol is computed from the current input and a set number of previous symbols. By spreading information across neighboring symbols, convolutional codes aim to preserve context in the stream and cope with the loss of codeword boundaries on channels that may delete bits.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Documents