Hidden Transmitter Syndrome
A Split Personality

Introduction

Hidden Transmitter Syndrome (also known as Hidden Terminal Syndrome or HTS for short) has been an important factor in packet radio communications from the beginning. Tadd (KA2DEW) and the "gang" have continually preached the necessity of HTS Free communication links as the answer to a really efficient network. NEDA has proved in its network that an HTS Free linked network really does work, usually by linking with "dedicated point to point links". However, a more difficult problem arises with user ports. Network managers have little or no control over the way users communicate with the user ports other than trying to educate them in the finer technical points of a CSMA environment.

Several suggestions have been promoted to reduce the HTS effects on user ports, the most common being "cellular LANs" and digital repeaters. I personally have operated a digital repeater for more than 4 years and have a good feel for their advantages and disadvantages. One person here in Canada has proposed that digital repeaters are the solution to the HTS problem on user ports. He suggested that all the HTS effects would disappear, and that up to 20 users would be able to use the port at the same time. He also indicated that the users would of course continue to use the radios they had available but everybody would have to convert to 9600 bps operation. He dismissed cellular LANs because "there are not enough high buildings to put all the cells on". This was an interesting excuse considering that it was his group that first promoted the idea of cellular LANs in amateur packet radio. These proposals got me thinking about the whole HTS question and why my own results with a repeater did not indicate a complete solution to the problem.

From my experience with a digital repeater, I had noted that during times of very high duty cycle operation (80% +) the throughput would be good at times with low retries and at other times one would need many more retries to get the data across. I knew that we had 6 heavy users on the repeater (which was a controlled backbone system). Four of the six users were using fast, crystal controlled radios, one had a synthesized radio and one was using closed squelch operation. Study showed that throughput was good when all the stations on the repeater at one time had fast radios but whenever either of the slow radio users were on, retries went up 3 or 4 times. These facts got me to studying Hidden Transmitter Syndrome in much greater detail. After much study and thought, I came to the conclusion that Hidden Transmitter Syndrome was really made up of two very different parts. To distinguish between them, I coined the terms "True HTS" and "Key-up HTS".

True HTS is the same hidden transmitter syndrome that has been blamed for all our problems since the beginning of time (packet time that is). It is due to the geographical separation or obstruction between stations that prevents one station from hearing another. CSMA (carrier sense, multiple access) environments rely on the "sensing of a carrier" to control access and hold-off transmit. If the carrier cannot be sensed, the CSMA channel is out of control. True HTS can be eliminated by reducing the geographical separation (cellular LANs) or using a "zero delay" repeater on the obstruction or other high point to relay the "presence of a carrier" to all stations. Condition solved! Or so one might think. Enter "Key-up HTS".

Key-up HTS is the hidden transmitter syndrome that is caused by:

• the delay between the keying of the PTT line by your TNC and the actual appearance of a radio signal on the channel plus

• (If applicable) the key-up and propagation delays in a digital repeater if used plus

• the delay from when a signal appears on the channel until the other station's receiver starts to detect it and output a demodulated data signal plus

• the delay from when the receiver outputs a demodulated data signal until the TNC recognizes it as a valid data signal and turns on the DCD line.

Until the other station's TNC "senses" your transmitted signal, you are a hidden transmitter even to your next door neighbor. However once your TNC decides to transmit, it is committed. If any other station decides to transmit during the Key-up HTS period when it did not know that you had started, a collision will occur. The result is just the same as if you had been on the other side of the state. Only the cause is different. What is the cure? The ONLY cure to Key-up HTS is the use of extremely fast radios by ALL stations on the channel or a different operating protocol (such as DAMA or other "master/slave" protocol).

but you say "i've got a fast radio. It's got "electronic switching". This doesn't apply to me." Well, I have news for you. First, the key-up hts is governed by the slowest radios on the channel, not just your radio. Secondly, it is not just the electronic switching that makes a fast radio, it is also the time for the frequency to stabilize and the power amplifier to come up in power as well as the recovery time of the receiver after a transmit period. During the past year, I have been developing a set of procedures for testing data radios. I have called it the Data Radio Standard Test Method (DRSTM) project. The DRSTM project has concentrated on the various radio delays by separating them into the smallest different segments. The results of tests on available radios have been very enlightening. For example, all the Icom synthesized radios I tested had an initial power-on delay of about 100 milliseconds.

"100 ms you say. That's nothing. Just a blink of the eye." Well, it may be a blink of the eye to a human but it is a VERY significant time delay to a TNC. For example, take a look at the default parameters in some of the TNCs being sold today. A recent PacComm Tiny-2 that I bought, had default persistence = 200 and slottime = 1 (10ms). If a station using this TNC, without changing its parameters, had data to transmit, then the chance that it would transmit within 100 ms. is 99.99994%. If another station had also decided to transmit during this period, a collision would be virtually certain. Even if the persistence was 128 and slottime = 3 (30 ms), the chance of a collision would still be approximately 90%. And it is my impression that the majority of packeteers are running with default parameters in their TNCs.

"So what is a fast radio?" A really fast radio is any radio whose maximum important time delays on both transmit and receive are less than the typical delays in a TNC (typically 10 ms). The important time delays are the time from the PTT line being keyed until the signal appears on channel (not necessarily stabilized) and the time from the release of the PTT line until the receiver can detect a signal on the channel. Crystal controlled radios can meet this requirement easily. On the other hand, the only phase locked radios that come anywhere near this requirement are radios using mixers to generate the output frequency. Multimode (ssb,cw,etc) radios use this mixing technique but almost none of the FM voice synthesized radios use mixing. Most are abysmally slow. They don't have to be, it's all in the design and until now there has been no incentive to design fast radios. For more details on testing data radios, contact the author for information on his DRSTM test manual.

"Well, the old relay switched radios are really slow, aren't they?" Don't believe it. Contrary to what has been published in many packet books and spread by many self-styled "packet gurus", most relays will switch in 20 ms or less. If a relay switched radio is slow in switching, it will much more likely be due to circuit design (eg receiver circuits might have delayed turn-on due to filter capacitors charging) than the relay speed. I recently tested a Motorola MOCOM 70 which has a big clunking relay for power switching. The transmitter turned on and was stabilized in 15 ms. The receiver recovery (at the detector) tested somewhat slower (60 ms) until it was found that a spike suppression diode across the relay coil was holding the relay closed after keying release. Replacing the diode with another suppression technique, lowered the receive recovery to 13-17 ms. So relays are not the villains that they are made out to be.

"What about packet ready or 9600 bps ready radios?". A packet ready or 9600 bps ready radio just means that it has a 6 pin connector with direct connections to the modulator and discriminator. Unless proven otherwise, assume that they are just as bad as the basic radio. If they are multimode radios, they will be reasonably fast. If FM voice, they will probably be slow. The only way to be sure is to test them. If my DRSTM project gets widely adopted, there will be a database of test results available. Until then, you will have to do your own tests.

"What about other protocols?" CSMA (carrier sense, multiple access) is a really poor choice for a shared, switched radio environment. But in North America, we seem to stick to it like a kid with an old blanket. Any of the "master/slave" protocols would be better. One protocol that seems to have received a lot of press is DAMA (demand allocation, multiple access). With DAMA, a master station polls the users (called slaves) and gives them each a chance to transmit. The master assigns a number of slots in the queue to each slave according to the amount of data that has been exchanged recently. The only chance for a collision is when a new user wishes to join the queue. Otherwise radio switching speed and other factors cannot affect the collision rate. DAMA does require new firmware in the user's TNC, so this has to be a factor in adopting it. DAMA is widely used by the commercial packet radio industry and also by the amateur packet people in Europe. It is worth a look and some of you might consider setting up a specialized DAMA user port to experiment with.

So to conclude, when planning packet networks and user interfaces in the future, both aspects of the Hidden Transmitter Syndrome must be considered. Especially if you are thinking of setting up digital repeaters or looking at increased loading of user ports at higher speed. If you do not consider the effects of Key-up HTS, you may be disappointed in the results of your efforts.