Dale R. Worley writes: > "Pascal Thubert (pthubert)" <[email protected]> writes: > > Seems that large EBs at low rates ended up being a problem in subgig > > cases after all, Dale. > > Ouch! > > My memory is that 802.15.4 limits the EB to 127 bytes (plus the > link-layer overheads), and that maximum-length frames can be transmitted > in the default slot size.
IEEE Std. 802.15.4-2015 EB can be up to any length that the PHY will support. In the normal 2.4GHz O-QPSK PHY this is 127 bytes. Other PHYs do have other limits, and for example the 802.15.4t "Amendment: Higher Rate (2 Mb/s) Physical (PHY) Layer" will support max frame size of 2048 bytes on the same 2.4GHz band (with higher data rate). 802.15.4t is now in sponsor ballot phase, so it might be out quite soon. > And that 802.15.4 fixes the EB format in a way that does not provide > for fragmentation. So I don't see how you could even transmit an > extra-long EB, much less have to worry about a slot for it. All of that will depend on the PHY you plan to use. > But I suspect that my inexperience is making me overlook how extra-long > frames can be handled. > > I can see a way of avoiding this by partitioning the optional IEs you > want to send in the EB into several subsets, transmitting first an EB > with subset 1, then an EB with subset 2, etc. I suppose you'd need yet > another IE to tell the listener which subset this is and how many > subsets there are so that the listener could know when it had heard the > whole set. In the 802.15.10 " Recommended Practice for Routing Packets in IEEE Std 802.15.4 Dynamically Changing Wireless Networks" which provides mesh networking and routing to the 802.15.4 they will include several IEs in EBs, and EBs are separated so that some EBs have TC IEs, some have NLM IEs and so on. And because NLM IEs can still be too large, they will send them in pieces, i.e., the NLM IE contains internal subsetting, where they send the whole list of neighbors over multiple NLM IEs. -- [email protected] _______________________________________________ 6tisch mailing list [email protected] https://www.ietf.org/mailman/listinfo/6tisch
