(I guess Andrew Blumberg moved from Stanford to the University of Texas.)
There might be an inherent tradeoff where you need at least one of {tamper-resistant trusted meters, at least slightly noisy measurements, potential deanonymization}. For example, the short paper mentions that "point tolls" are easy to make anonymous using any form of anonymous digital cash (or blinded tokens issued by the tolling authority!), but the exact usage billing you mention people wanting is much more detailed than a point toll like that. It might indeed be inherently impossible to get all the way there without detailed surveillance.
https://www.eff.org/files/eff-locational-privacy.pdf (2009)
The technical paper mentioned is now at
https://web.ma.utexas.edu/users/blumberg/vpriv.pdf
(I guess Andrew Blumberg moved from Stanford to the University of Texas.)
There might be an inherent tradeoff where you need at least one of {tamper-resistant trusted meters, at least slightly noisy measurements, potential deanonymization}. For example, the short paper mentions that "point tolls" are easy to make anonymous using any form of anonymous digital cash (or blinded tokens issued by the tolling authority!), but the exact usage billing you mention people wanting is much more detailed than a point toll like that. It might indeed be inherently impossible to get all the way there without detailed surveillance.