We probe the limits, both experimentally and analytically, of passive optical network (PON) monitoring using periodic coding technology. The experimental demonstration focuses on a 16 customer PON with a 20 km feeder fiber followed by either a single cluster or a tiered hierarchy. A directly modulated laser modulated at 1 GHz was used to generate the monitoring probe signals. The measured data from the experimental setup was fed to a reduced complexity maximum likelihood sequence estimation (RC-MLSE) algorithm to detect and localize the customers. Three different PON deployments were tested. We demonstrate improved monitoring robustness when using a variable threshold for networks with a tiered geographic distribution. While only a 16 customer PON was tested, our experimental setup had 18 dB margin in the total loss budget corresponding to splitting losses for 64 customers. We investigate analytically the total permissible loss budget of the monitoring system operating in the 1650 nm waveband as a function of receiver specifications. We examine the effect of resolution in the analog-to-digital conversion on the correlation peaks that form sufficient statistics for the RC-MLSE algorithm. Resolution affects both the RC-MLSE algorithm and the use of signal averaging to improve signal-to-noise ratio. We find that the monitoring system is able to monitor current PON standards with inexpensive, commercially available electronics.
© 2011 IEEEPDF Article