Slow frequency hopping - coordination

phm
Neuling
Neuling
Beiträge: 8
Registriert: 6. Mai 2009 18:55

Slow frequency hopping - coordination

Beitrag von phm » 12. Mär 2010 11:42

There's something I don't really understand about slow FHSS. First, here's what I've understood about SS techniques (to check if it's correct):

Spread Spectrum techniques (or CDMA) are used to allow several users to communicate at the same time on the same channel, in a more flexible way than TDMA and FDMA (and SDMA), and with little coordination effort.

With DSSS, a sender S multiplies a signal by a chipping sequence before transmitting it, and the same sequence is used at the receiver R to extract the signal from the channel. Although other signals are transmitted at the same time on the same channel and with the same power, they will be perceived as noise by R since they were multiplied by another sequence. Thus the signal can be correctly recovered. However, if too many users transmit at the same time, the noise level will be too high and errors will start to happen frequently, that's why users have to be dropped out of a CDM cell (cell breathing).

Fast Frequency Fopping has a similar result to DSSS (although less robust).

Things are different for slow hopping: the overall spectrum is spread, but for the time of a packet transmission only a narrow frequency band is used. It is similar to FDMA except that the frequencies change for each packet/slot, and there is no (or little) coordination between the senders/receivers.


Now, my question:
Shouldn't there be more coordination for slow FHSS than for DSSS and fast FHSS? For the latter ones, the fact that the spreading sequences of 2 pairs of sender/receiver overlap doesn't matter for a correct decoding. But in case of slow FHSS, if 2 senders use the same frequency for the same slot, both packets will get lost.
For Bluetooth, the hopping pattern is determined by the device ID of the master, which is unique worldwide (so I guess it's fix for the lifetime of a node). Does this mean that the transmissions of 2 co-located piconets with overlapping hopping patterns will systematically interfer in some given slots? And would be the only solution to systematically retransmit packets transmitted in these slots?

Matthias Hollick
Dozentin/Dozent
Beiträge: 99
Registriert: 21. Okt 2005 15:42

Re: Slow frequency hopping - coordination

Beitrag von Matthias Hollick » 14. Mär 2010 18:45

My answer goes inline. Please also note that we discussed the PHY aspects of wireless communications mainly to get a grip on the design of higher layer protocols; hence PHY-related topics might not be the topics the exam focuses on.
phm hat geschrieben:There's something I don't really understand about slow FHSS. First, here's what I've understood about SS techniques (to check if it's correct):

Spread Spectrum techniques (or CDMA) are used to allow several users to communicate at the same time on the same channel, in a more flexible way than TDMA and FDMA (and SDMA), and with little coordination effort.
This is correct. Given these users use different spreading codes/sequences, they can use the channel at the same time. However, there is still need for coordination (to let the receiver know about the spreading sequence or hopping pattern). Spread spectrum also serves to make communications more robust against narrowband interferes; it was historically used to increase the "confidentiality" of the transmission (by "hiding" the transmission, since the signal was harder to detect).
phm hat geschrieben: With DSSS, a sender S multiplies a signal by a chipping sequence before transmitting it, and the same sequence is used at the receiver R to extract the signal from the channel. Although other signals are transmitted at the same time on the same channel and with the same power, they will be perceived as noise by R since they were multiplied by another sequence. Thus the signal can be correctly recovered. However, if too many users transmit at the same time, the noise level will be too high and errors will start to happen frequently, that's why users have to be dropped out of a CDM cell (cell breathing).
Exactly.
phm hat geschrieben: Fast Frequency Fopping has a similar result to DSSS (although less robust).
The difference between fast hopping and slow hopping is mainly that a slow hopping system changes ("hops") frequencies only after transmitting a number of bits, while a fast hopping system change frequencies much more often (multiple times during the transmission of a bit). The signal is sort of spread in the time domain, while the channel frequency is kept constant only for short (very short) amounts of time. As a result FHSS has similar results as DSSS, but since it does not spread over the entire bandwidth, but uses one narrow channel at a time, it is less robust than DSSS.
phm hat geschrieben: Things are different for slow hopping: the overall spectrum is spread, but for the time of a packet transmission only a narrow frequency band is used. It is similar to FDMA except that the frequencies change for each packet/slot, and there is no (or little) coordination between the senders/receivers.
See above. Fast hopping is like slow hopping, but since it hops multiple times per bit transmitted it has different characteristics.
phm hat geschrieben: Now, my question:
Shouldn't there be more coordination for slow FHSS than for DSSS and fast FHSS? For the latter ones, the fact that the spreading sequences of 2 pairs of sender/receiver overlap doesn't matter for a correct decoding. But in case of slow FHSS, if 2 senders use the same frequency for the same slot, both packets will get lost.
If for DSSS or fast FHSS the difference in spreading sequences is too "small" you will have very high error rates. But you are right: you damage only "parts" of bits, while in s slow FHSS, a number of bits is destroyed if two or more devices use the same frequency at the same time. To be more robust protocols such as Bluetooth employ adaptive FHSS, which tries to identify channels that are prone to interference and "blacklists" these channels, i.e., it excludes these from its hopping pattern.
phm hat geschrieben: For Bluetooth, the hopping pattern is determined by the device ID of the master, which is unique worldwide (so I guess it's fix for the lifetime of a node). Does this mean that the transmissions of 2 co-located piconets with overlapping hopping patterns will systematically interfer in some given slots? And would be the only solution to systematically retransmit packets transmitted in these slots?
This is a good question. Honestly, out of the top of my mind, I cannot give you the authoritative answer. However, it is obvious that thew different piconets cannot operate collision free, but will collide in a number of slots. For usual operating conditions the employed forward error correction of bluetooth should take care of this. I am sure that there is some literature where researchers studied the amount of interference between co-located random bluetooth piconets, though ... you might want to check it out to find the correct answer to your last question.
Matthias Hollick

matthias.hollick (at) seemoo.tu-darmstadt.de

Antworten

Zurück zu „Archiv“