tRTS packet by returning a CTS and entering a- 2 -ight read loop, waiting for the data to arrive.
n(A timeout prevents a system lockup if the dataever arrives.)As is standard for CSMA schemes,nCSMA/CA requires stations to stay off the chan-el when another transmission is already in pro-togress. CSMA/CA also requires any station thaverhears an RTS or CTS packet directed else-twhere to inhibit its transmitter for a speciﬁedime. This helps reduce the probability of a col-.Tlision with a subsequent CTS or data packethis is the CA or
part of rpCSMA/CA. However, collisions are not a majoroblem on Localtalk; the network is physicallyarsmall, carrier sensing is fairly rapid, the datate is relatively low, and (if the network is.Pproperly built) there are no hidden terminalslain CSMA would work well, but there was lit-Rtle extra cost to the CA feature (given that theTS/CTS dialogue was already needed for other
reasons) so it was included.
. Turning CSMA/CA into MACA
nsHidden and exposed terminals abound oimplex packet radio channels, and this makestothem very different from Localtalk and mosther types of local area networks. When hid-taden terminals exist, lack of carrier doesn’lways mean it’s OK to transmit. Conversely,rwhen exposed terminals exist, presence of car-ier doesn’t always mean that it’s bad totltransmit. In other words, the data carrier detecine from your modem is often useless. So I’llomake a radical proposal: let’s ignore DCD! Inther words, let’s get rid of the CS incCSMA/CA. (It’s too hard to build good DCDircuits anyway...)Instead we’ll extend the CA part of whatk we’ll call MA/CA (or just plain MACA). Theey to collision avoidance is the effect that RTSntand CTS packets have on the other stations ohe channel. When a station overhears an RTSntaddressed to another station, it inhibits its owransmitter long enough for the addressed stationn
to respond with a CTS. When a statio
It would be nice if we could use this feature onp
acket radio with our programmed-I/O HDLC interfaces,i(e.g., DRSI PCPA, Paccomm PC-100). Unfortunatelyf our RTS/CTS packets carry full source and destina-Ftion call signs, they would not ﬁt into the 3-byte 8530IFOs. So high speed operation will still require eitherDMA or a dedicated I/O processor.
overhears a CTS addressed to another station, itoinhibits its own transmitter long enough for thether station to send its data. The transmitter isshinhibited for the proper time even if nothing ieard in response to an RTS or CTS packet.ZcFigure 3 shows an example. Stationannot hear X’s transmissions to Y, but it
aChear Y’s CTS packets to X. If Z overhearsTS packet from Y to X, it will know not toXtransmit until after Y has received its data from.But how does Z know how long to waithafter overhearing Y’s CTS? That’s easy. Weave X, the initiator of the dialogue, include inosits RTS packet the amount of data it plans tend, and we have Y, the responder, echo thatoinformation in its CTS packet. Now everyoneverhearing Y’s CTS knows just how long totmwait to avoid clobbering a data packet that iight not even hear.As long as the link between each pair of sstations in the network is reciprocal (i.e., all thetations have comparable transmitter powers andpreceiver noise levels), the receipt of a CTSacket by a station not party to a dialogue tellsiit that if it were to transmit, it would probablynterfere with the reception of data by thesiresponder (the sender of the CTS). MACA thunhibits transmission when ordinary CSMAsrwould permit it (and allow a collision), thuelieving the hidden terminal problem. (Colli-tlsions are not
avoided; more on this poinater.)Conversely, if a station hears no responsettto an overheard RTS, then it may assume thahe intended recipient of the RTS is either downe4or out of range. An example is shown in ﬁgur. Station X is within range of Y, but not Z.pWhen Y sends trafﬁc to Z, X will hear Y’s RTSackets but not Z’s CTS responses. X mayf itherefore transmit on the channel without fear onterfering with Y’s data transmissions to Z,Meven though it can hear them. In this caseACA allows a transmission to proceed when,tordinary CSMA would prevent it unnecessarilyhus relieving the exposed terminal problem.a(Because modems have a capture effect, hearingCTS doesn’t
mean that you’d cause atycollision if you transmit, so the problem isn’et completely solved. More on this point later.)