-based noise suppression in spectrum sliced PONs: impact of bit rate and gain recovery time F. Vacondio, W.Mathlouthi,, P. Lemieux and L. A. Rusch Centre d optique d photonique et laser (COPL) Department of Electrical and Computer Engineering, Université Laval IASTED Wireless and Optical Communications (WOC) Montréal, June 1st 27 Outline -1- -2- Motivation gain recovery time Noise suppression experimental setup Results Conclusion
SS WDM PON Scenario Schematic of a passive optical network (PON) with incoherent shared light source: Data Optical Network Unit () Optical Network Unit () Incoherent Broadband Source AWG MZ Optical Line Terminal (OLT) M U X feeder M U X Beat noise of incoherent light -3- -4- With incoherent light, the SNR is B proportional to o B e High intensity noise can cause BER floors, and there is a tradeoff between bit rate and spectral efficiency
based noise suppression If we use a Intensity noise is heavily reduced. Experimental Setup -5- -6- We focus on two proposed schemes for noise suppression: (a) Broadband source.24 nm BPF1 PC Data MZ Current BPF2 Detector Booster configuration (with and without BPF2) (b) Broadband source BPF1 PC Data MZ OLT BPF2 Current BPF3 Detector CW 5/5 Conversion configuration (incoherent to coherent, XGM) We will investigate the influence of gain recovery dynamics of s, by using two different devices. (Booster will be measured with and without BPF2 notice that avoiding it is unrealistic)
gain recovery time Normalized Gain Recovery 1.8.6.4.2 1 2 1 2 3 4 5 6 7 Time [ps] 1 (CIP NL OEC 155) recovers the gain in 25 ps 2 (Covega BOA 2679) needs 65 ps (1% 9% time). ASE dynamics (I) -7- -8- Fast : Important ASE modulation. The ASE degrades both Extinction Ratio and Eye Opening Slow : ASE is quasi CW signal Less Eye opening penalty
ASE dynamics (II) -9- Normalized modulation response [db] Slower 2 has the a smaller ASE modulation bandwidth, and 2mW total ASE power. Faster 1 has the large ASE modulation bandwidth ~ 4GHz (and 11mW total ASE power) Results -1-1 1-2 Booster configuration at OLT 1.25 Gbps Conversion configuration at No BER floor induced by the impact of the ASE on the extinction ratio for the faster 1. BER 1-4 1-6 1 (fast) 2 (slow) 1 (fast) 2 (slow) No ber floor and 1 is more effective, since XGM is more efficient for fast s. 1-8 1-1 -4-35 -3-25 -2-15 -1-5 P in [dbm] Input power to. Recall these are back to back measurements
no filter Results Conversion Conversion (XGM) 1 11 1 1-2 Conversion configuration at 1 (fast) 5 Gb/s 1-4 BER 1-6 2.5 Gb/s 1-8 622 Mb/s 1-1 1.25 Gb/s -4-35 -3-25 -2-15 -1-5 P [dbm] in Independently from BR, fast 1 is more effective than 2. (no floors up to 2.5 Gbps) Results Booster -11- -12- filtered 1-2 Booster configuration at OLT 2 (slow) 5 Gb/s BER no filter 1-4 filtered 1-6 2.5 Gb/s 622 Mb/s 1-8 1.25 Gb/s 1-1 -4-35 -3-25 -2-15 -1-5 P [dbm] in If BPF2 is taken away, we are detecting all ASE. Especially at BRs at which its ASE cannot follow, slower 2 is more effective!
Conclusions For conversion configuration fast s are more effective (XGM efficiency). Advantage: for bit rates as high as 2.5Gb/s, we were able to eliminate BER floors. Disadvantage: the need to saturate the at the receiver so that XGM is efficient. For booster configuration slow s are more effective (ASE dynamics). Advantage: the possibility of using the as a booster. Disadvantage: for bit rates as high as 2.5 Gb/s, BER floors appear. -13- -14- Thank You for your kind attention!
Gain recovery setup Normalized Gain Recovery 1.8.6.4.2 1 2 1 2 3 4 5 6 7 Time [ps] (a) (b) Ase Mod. Setup -15- -16-