Content area
Full text
Abstract. A new variant of the selective-repeat protocol is presented. While the protocols used until now have been based on cumulative acknowledgments and a single retransmit timer, a protocol based on individual acknowledgments and a separate timer for each outstanding packet is proposed here. In spite of a slightly increased complexity of the proposed protocol, when comparing it to the basic selective-repeat protocol, it is still relatively simple to be implemented; furthermore, it seems to be even less complex than the selective-reject protocol. The simulation results presented in the paper show that the new protocol is more efficient not only than the basic selective-repeat protocol, but also than the selective-reject protocol. Because the basic protocol is especially nonefficient when several packets in line are lost, the advantage of the proposed protocol is especially distinctive in case of heavy losses in the channel, but also in case of long channel delays. The relative efficiency of the proposed protocol is almost the same as the relative efficiencies of the basic and selective-reject protocols. Therefore, the new protocol seems to be most suitable in nonmultiplexing environments, especially in the data-link layer where cumulative acknowledgment single timer go-back-N protocols have mostly been used until now.
Keywords: selective-repeat protocol, selective-reject protocol, retransmit timer, individual acknowledgment, protocol efficiency
Protokol s selektivnim ponavljanjem ter vec casovniki za ponovno oddajo in individualnimi potrditvami
V clanku je predstavljena nova razlicica protokola s selektivnim ponavljanjem. Medtem ko smo doslej uporabljali protokol s kumulativnimi potrditvami in enim samim casovnikom za ponovno oddajo, tu predlagamo uporabo protokola z individualnimi potrditvami in posebnim casovnikom za vsako ze oddano, a se ne potrjeno sporocilo. Ceprav je predlagani protokol nekoliko kompleksnejsi kot osnovni protokol, ga je danes vseeno mogoce dovolj preprosto implementirati; kaze pa, da je predlagani protokol celo manj kompleksen kot protokol z negativnimi potrditvami. Rezultati simulacij kazejo, da je tak protokol ucinkovitejsi ne le od osnovnega protokola s selektivnim ponavljanjem, ampak tudi od protokola, ki poleg pozitivnih uporablja tudi negativne potrditve. Ker osnovni protokol deluje neucinkovito predvsem takrat, ko se zaporedoma izgubi vec sporocil, je prednost predlaganega protokola se zlasti izrazita pri velikih pogostnostih izgub, pa tudi pri velikih zakasnitvah v kanalu. Relativna ucinkovitost predlaganega protokola pa je priblizno enaka kot relativna ucinkovitost osnovnega protokola in protokola...