On Tue, Sep 7, 2010 at 3:34 PM, Carsten Bormann <[email protected]> wrote: >> Since 15.4g has defined a MTU of 2047, is there still a need to use >> 6lowpan-hc there? > > I would very much think so. > > While reducing fragmentation is an important motivator for 6lowpan-hc in > 802.15.4 networks, the significant packet size reductions are useful in any > constrained network, as every byte actually sent and received: > > -- consumes power at both sender and receiver, > -- increases the packet transmission time and this makes it subject to > additional bursts of interference, reducing delivery probability, > -- increases contention of the channel (which, in turn, also may cause > interference). > > With the large headers used by IPv6 and the small payloads expected for > 6lowpan applications (e.g., CORE's CoAP protocol), the ~30-40 byte per packet > savings will be worth it in these environments even if the hazards of > MTU-related fragmentation are going away.
I remember that in RFC4944 it states the small MTU is the main motivation for header compression. The reasons you stated above are reasonable only if the complexity incurred are considerable low. I agree that 6lowpan is not designed for a certain MAC/PHY, otherwise it couldn't be used long since the MAC and PHY are evolving. Thank you, Zhen _______________________________________________ 6lowpan mailing list [email protected] https://www.ietf.org/mailman/listinfo/6lowpan
