The Suspected Frsky GUID Issue

Problems, experiences or just something to say about RC gear? Say it here.

The Suspected Frsky GUID Issue

Postby aesmith » Thu Dec 02, 2010 9:25 am

Hi,

I've just been reading the various comments, and a couple of thinks puzzle me. One comment refers to the fail-safe kicking in and providing some damage limitation. That surprised me a little, because it suggests that the receivers detected that something was wrong. How would they do that?

The other was the advice to check transmitters before switching them on. What's the best way to do this?
aesmith
 
Posts: 101
Joined: Tue May 04, 2010 11:45 am
Location: Scotland

Re: The Suspected Frsky GUID Issue

Postby Ken Henderson » Thu Dec 02, 2010 6:02 pm

If FRSky are using randomly assigned 16 bit numbers as TX GUID´s then I am shocked at their un-professionalism and wish I had never invested in their equipment! The correct way to engineer this kind of system is to allocate unique GUID´s to each TX and I sincerely hope that there has been a misunderstanding somewhere along the line and that the problem lies somewhere else.

Waiting & watching in anticipation.....
Ken Henderson
 
Posts: 3
Joined: Thu Dec 02, 2010 5:48 pm

Re: The Suspected Frsky GUID Issue

Postby RCModelReviews » Thu Dec 02, 2010 8:05 pm

FrSky are investigating the situation and the affected gear will be returned to them for examination.

Given the amount of FrSky in use and the fact that this is the only reported incident I'm not overly concerned at this point but it is something to be aware of. I note similar problems were reportedly experienced by Corona users in Europe recently.

It could be that this is just a "lottery event" where, despite the tiny odds, something that is very unlikely, actually happened.

The odds that it would happen *here* so that *I* could observe it are even more remote.

As I said in the article, I'll be keeping a close eye on what is discovered.

FrSky's latest response is "we think this is a small probability event with very slim possibility".

I agree -- but still think it important that when FrSky users are flying together for the first time, they check for any possible interactions.
RCModelReviews.com, just the facts.
User avatar
RCModelReviews
 
Posts: 2120
Joined: Tue May 04, 2010 3:40 am

Re: The Suspected Frsky GUID Issue

Postby Foxbait1 » Thu Dec 02, 2010 8:49 pm

wish I had never invested in their equipment!
You could send the frsky equipment my way an go back to (or try) dsm2, no problems in that camp. lol Joking a-side got to agree with xjet. Any system that assignes a random guid may have a slim chance of another with the same guid. If there is one with the same guid as your`s good chance they`re half a world a-part an never see each other.
Foxbait1
 
Posts: 8
Joined: Tue May 04, 2010 5:35 pm

Re: The Suspected Frsky GUID Issue

Postby Ken Henderson » Thu Dec 02, 2010 9:52 pm

Perhaps someone with a better statistical brain could give some examples of the actual probabilities of two identical GUID´s occurring within a 16 bit number sample (65536 decimal possibilities). The mathematics look to me to be the same as discussed in the "birthday problem" see:-

http://en.wikipedia.org/wiki/Birthday_problem

One figure I have calculated (perhaps!) is that it would take a population of 302 TX´s to give a better than 50% probability of two being on the same GUID. My club has 160 members (fortunately not all using FRSKY) but if they were, I suspect that chance that two of us could be on the same GUID would be statistically low(ish) but still uncomfortable. All of this could be avoided if a larger number of bits were used and unique assignments made.....

Any mathematically inclined takers....
Ken Henderson
 
Posts: 3
Joined: Thu Dec 02, 2010 5:48 pm

Re: The Suspected Frsky GUID Issue

Postby RCModelReviews » Thu Dec 02, 2010 10:09 pm

This is a Birthday Problem issue to a degree -- if you assume that all the radios are on simultaneously.

When you look more closely and realize that probably, even at a busy meeting, no more than a dozen radios would be in use simultaneously, the odds become a little longer that a clash will occur.

There's also the fact that our radios are also separated not only by GUID but also by frequency. In the case of a FHSS system, they'll only interfere when they happen to hit the same frequency at the same time. Remember the old-days when we had people with *NO* GUIDs flying together and separated only by frequency on bands like 35/36/40/72MHz?

Unfortunately, it appears that the hopping sequence for the two radios involved in this clash seemed to drift in and out of sync. This explains why both models were able to take off and fly for a while before being affected. It also explains why, when left turned on, the interaction between the radios would "come and go" -- as the hopping sequences coincided and then drifted apart.

I've suggested that, at least as an interim measure, FrSky print the GUID on the back of each module and DIY board. This would allow people to easily confirm that they didn't have any conflicts at their local flying field. They may also want to release a piece of software or a small piece of hardware that allows the GUID of any existing module/DIY to be read so that people can annotate their existing equipment.

This is an unfolding story and I'll keep everyone updated with more info as it comes to hand.

Still no reports from elsewhere of FrSky systems clashing so it is increasingly looking like a "lottery event".
RCModelReviews.com, just the facts.
User avatar
RCModelReviews
 
Posts: 2120
Joined: Tue May 04, 2010 3:40 am

Re: The Suspected Frsky GUID Issue

Postby Heather » Fri Dec 03, 2010 12:31 am

Computer based random number generators are Psuedo random so may not fit all statistical models. Of course if you don't start with a random seed value you will get the same sequence of numbers every time.

Heather
Heather
 
Posts: 12
Joined: Thu May 27, 2010 9:40 am

Re: The Suspected Frsky GUID Issue

Postby kneedrag » Fri Dec 03, 2010 12:58 am

To be clear this is the first reported issue of this happening. I will not mention another large brand name manufacture who's tx's all defaulted there GUID's to zero's effectively allowing any of there tx's to take control of another rx.

Granted this should not have happened it the first place however FrSky is addressing the issue on the first occurrence of this. Not many manufactures address issues like this on the first occurrence.

No system is perfect as much as we would like them to be.

However until FrSky comes out with an official fix it would pay just to be on the safe side and ground check your models before flying if there are multiple FrSky users about especially with new club members or people who have new tx systems.

I have a more detailed thread which FrSky is watching and will take part in if you have any questions around this issue. http://www.parkflyers.org.nz/modules/ne ... mpost28466

Further to note not only would the system have to be on the same GUID for this to happen but on the same set of Freq channels. Due to this being a large flyin and the Spectrum range was fairly full it would not have helped them finding a free spot in the band and would have compounded the issue. If it was just the two of them flying we would most probably never have seen the issue, that would have been far worse as the issue would not have been seen and not addressed.
kneedrag
 
Posts: 105
Joined: Thu May 06, 2010 9:44 am

Re: The Suspected Frsky GUID Issue

Postby Roger » Fri Dec 03, 2010 2:32 am

Hey Rob / Bruce, I wonder if this issue was also affecting my Corona system at the weekend too? The little receiver in my Alula handlaunch went nuts for a while and wouldn't bind up properly - yet when I got it home there was nothing wrong with it... yet I had the same type of receiver in my Chucko and it flew all weekend with no problems - except from my terrible last landing! :(
Roger
 
Posts: 79
Joined: Mon May 10, 2010 11:00 am

Re: The Suspected Frsky GUID Issue

Postby gruvin » Fri Dec 03, 2010 3:21 am

Assuming you meant that you were trying to bind or re-bind the receiver at the flyin, as opposed to just switching it on and not getting a link then,...

I don't know exactly how the various binding protocols work. But I wouldn't mind betting that one would be pushing their luck trying to (initially) bind a single given transmitter/receiver pair in the presence of many other signals. Once bound, perhaps necessarily in a 'quieter' place, things should be fine though of course. If another transmitter is also in bind mode and in range ... well I'm not sure what would happen.

Anyway, my simple point is that I think initial binding (pairing) of receivers may need to be done in a fairly RF-quiet place -- or at least with the transmitter very close. However, too close can cause problems I've found. For example, my Spektrum AR7000 receivers jittering servos and have some difficulty binding if inside a few centimetres of the transmitter. This is fair enough though and to be expected. My ow brain does not 'pair' so well with people screaming right in my ear either! :-P)
gruvin
 
Posts: 7
Joined: Tue Sep 21, 2010 10:16 pm

Next

Return to Radios and Servos

Who is online

Users browsing this forum: No registered users and 6 guests

cron