While millimeter-wave radio frequencies are hyped as the future of high-speed wireless networks, they're severely range-limited. A group of Spanish boffins has proposed using user context information like location to help mobile devices get the best speed.
As they explain in their paper at Arxiv, it's not enough just to run an efficient search for base stations: whatever algorithm is implemented, users also need fast switching between base stations.
As the paper, by Antonio Capone and Ilario Filippini of the Politecnico di Milano and the IMDEA Networks Institute's Vincenzo Sciancalepore, notes: “even with the massive use of sophisticated directive transmissions the availability of mm-wave access in future mobile networks cannot be continuous”.
Discontinuities will arise because of the limited coverage areas of mm-wave signals – even the human body is enough to obstruct propagation. Not only does this interrupt voice or data traffic, it's also a problem for signalling.
In the MiWEBA project, which kicked off in October 2014, the researchers propose splitting the control plane and the user plane. That way, control plane communications can be maintained over a lower frequency but more available channel (a conventional cell, for example), leaving the more fragile mm-wave communications to support user traffic.
As the paper notes, the kinds of base station search used in mobile telephony – all the way up to LTE – simply won't work fast enough to cope in a mm-wave world. In those, a device begins joining a network with a random search, and maintains its connection with an omnidirectional synch.
Neither of these work well under mm-wave constraints, the boffins write, so they propose giving the mobile device geographic knowledge of the network so its search is no longer random.
Their modelling suggests that not only would this outperform the random search: mobiles could remember signal behaviour in particular locations, so as to favour the best beam-forming configuration to get connected quickly. ®