Re: [6tisch] The channel hopping scheme seems to be suboptimal

Dale,

Please note that RFC7554 is informational, and describes/introduces the
functioning of TSCH. The actual standard is IEEE802.15.4-2015 [1]. In
particular, in case of discrepencies between the docs, IEEE802.15.4-2015 is
the one to follow.

You raise a valid point. There are a couple of points, though, that make an
"Absolute Slot Number" more favorable compared to an "Absolute Slotframe
Number":
- if you schedule multiple cells between two nodes in a single slotframe,
you want those different transmissions to happen at a different frequency
- ASN is used to construct a nonce when securing link-layer frames.
Security is such that we never want to re-use the same nonce.

Happy to answer further questions to might have, also during the interim
meeting later today [2].

Thomas

[1] http://standards.ieee.org/about/get/802/802.15.html
[2] https://www.ietf.org/mail-archive/web/6tisch/current/msg04760.html

On Thu, Sep 22, 2016 at 4:39 PM, Dale R. Worley <worley@ariadne.com> wrote:

> Pardon me, I'm new here, but I was wondering why the channel hopping
> scheme is defined the way it is.  Currently, RFC 7554 notes that the
> channel selected for a transmission by a node is:
>
>    frequency = frequency_table [ (ASN + channelOffset) mod nFreq ]
>
> where channelOffset is a set value for this node to use in this slot of
> the slotframe cycle, frequency_table is a set array of frequencies, and
> nFreq is the size of frequency_table.
>
> The question I have is with ASN, the "absolute slot number".  RFC 7554
> states that it increments one for each slot time, so that as a
> particular slotframe slot recurs, the ASN is incremented by the number
> of slots in the slotframe.
>
> The result is that to ensure that each node's transmissions cycle
> through all the frequencies in the frequency_table, the length of the
> slotframe needs to be relatively prime to nFreq:
>
>    A way of ensuring communication happens on all available frequencies
>    is to set the number of timeslots in a slotframe to a prime number.
>
> My question is that if ASN was redefined to be a counter of slotframes
> rather than of slots, then each node would cycle through all the
> frequencies regardless of the size of the slotframe, because each node's
> frequency index would advance 1 with each slotframe rather than
> (slotframe_size mod nFreq).
>
> This alternative is fairly obvious, so there must be a known reason why
> it isn't desirable.
>
> Dale
>
> _______________________________________________
> 6tisch mailing list
> 6tisch@ietf.org
> https://www.ietf.org/mailman/listinfo/6tisch
>

-- 
_______________________________________

Thomas Watteyne, PhD
Research Scientist & Innovator, Inria
Sr Networking Design Eng, Linear Tech
Founder & co-lead, UC Berkeley OpenWSN
Co-chair, IETF 6TiSCH

www.thomaswatteyne.com
_______________________________________