GreenTouch Plants a Greener PON
The consortium's goal is to reduce networking technologies' power consumption by a factor of 1,000, which is why it's dabbling in the "radical." GreenTouch is the same group that introduced the Large-Scale Antenna System about a year ago, using 100-antenna arrays for wireless service. (See GreenTouch Demos Efficient Antenna.)
In other words, the idea is to not play nicely within the boundaries of standards. "If you really want a factor of 1,000, then you really need a clean slate," says Peter Vetter, who was the principal investigator on Bi-PON.
Tuesday, the group announced its first wireline project, called bit-interleaved PON (Bi-PON).
PONs are passive, but a lot of energy gets eaten up at a connection's endpoint, particularly the side on the home. The Optical Network Unit (ONU) receives all of the traffic that came down that particular branch of the PON. It processes all of that traffic, and only at the last second (well, microsecond) does it discard the stuff that's not meant for that particular house.
Shouldn't there be a better way?
Consortium members including Bell Labs and IMEC decided to try a system where the ONU only receives the traffic intended for its owners, ignoring everything else. Bi-PON does this using time-division multiplexing (TDM): It organizes data into time slots, with the ONU picking up only the traffic allotted to its slot.
That might sound wasteful, too, but it's a flexible version of TDM. Only the ONUs that need to receive data are given time slots, and the periodicity of the time slots can be changed on the fly. The whole idea is to keep things streamlined.
So, how does Bi-PON's power consumption stack up? The power used by the ONU comes out to about 100 mW, compared with 2 W normally, GreenTouch says. (That's comparing GreenTouch's FPGA-based platform with an XG-PON ONU also built from FPGAs.)
Commercialization would be tricky, because the ONUs and the central office need to be talking to each other continually, in a more advanced dialogue than they have today. Any commercial version would be about a decade away, Vetter thinks.
— Craig Matsumoto, Managing Editor, Light Reading