reliable and feasible technology for more than one gigabit over standard SI-POF

PERFORMANCE RESULTS

Introduction

The simulation framework allows us to make performance predictions in order to dynamically and iteratively adapt the design along their different stages. This provides us and our partners a complete knowledge from the earliest to the last stage of the design, enabling regressions for each one.

The simulation results presented here are such that the full functionality of the communication system is enabled and evaluated, i.e. synchronization, timing-recovery, channel estimation, noise estimation, channel equalization, rate adaptation of channel coding sub-system, etc.

We only present some simulation scenarios that we have considered representatives to show the capabilities of KDPOF technology. Since KDPOF design is configurable and scalable it can be updated to be used from short reach to very long reach applications. For the first case the capacity is translated in very high data-rates (> 1 Gbps) over short POF links. However, for the last case the capacity is translated in a high coverage (long POF links) working with lower data-rate (hundreds of Mbps).

Test scenario

The results are reported for the next scenario:

1. • SI-POF: Mitsubishi Eska Premier GH4001 (1 mm, NA 0.5)
   

2. • Photo-diode and Trans-Impedance Amplifier (TIA): ηext = 0.75, UMD NA 1.0, CJ = 1.8pF, RSH = 50MΩ, RS = 176.4Ω, LS = 8nH, Idark = 1nA, Cf and Rf optimized, T = 25ºC (values from [1]), where:
   

3. • ηext is the quantum efficiency of photo-diode
   

4. • CJ is the p-n junction capacitance of photo-diode, which increases with silicon area
   

5. • RSH is the shunt resistance (~MΩ)
   

6. • RS and LS are the serial resistance and serial inductance
   

7. • Idark is the dark current of photo-diode, due to reverse voltage in photoconductive configuration
   

8. • Cf and Rf compose the feedback circuit of TIA, and are optimized for each test
   

9. • The noise model is composed by shot noise, quantum noise and thermal noise from photo-diode and thermal noise of all the resistances of TIA circuitry
   

10. • Commercial Op Amps are used to implement the TIA and differential and common mode input and output characteristics are included in tests, as well as input referred voltage and current noises
    

11. • Data-rate vs. length @ BER < 10-8
    

Different light emitters are considered:

1. • Red LED: 650 nm, Δλ = 30 nm, Pavg = -1 dBm, 1st-order cut-off frequency fc-3dB = 50MHz (values from several authors [1]). Launching UMD (Uniform Mode Distribution) NA = 0.3
   

2. • LD: 650 nm, Δλ = 2 nm, Pavg = 5.5 dBm, underdamped 2nd-order cut-off frequency fc-3dB = 3GHz, damping coefficient ς = 0.35 (values from several authors [1]). Launching GMD (Gaussian Mode Distribution) NA = 0.25
   

3. • Green LED: 520 nm, Δλ = 40 nm, Pavg = 1.5 dBm, 1st-order cut-off frequency fc-3dB = 75MHz (values from Nichia sample reported by [1]). Launching UMD NA = 0.3
   

4. • Blue LED: 470 nm, Δλ = 30 nm, Pavg = 3.0 dBm, 1st-order cut-off frequency fc-3dB = 75MHz (values from DieMount sample reported by [1]). Launching UMD NA = 0.3
   

Short reach applications

The system is configured to work with a symbol-rate of fS = 250 MHz. The transmitter uses a DAC of 12 bits and the receiver an ADC of 10 bits. Both components work at symbol-rate and they are available nowadays in the market. Next figure shows the performance as data-rate vs. standard SI-POF length for the different light emitters. Orange rectangles show capacity covered with current commercial products.

As can be seen in figure, the low attenuation of green and blue windows increases the data-rate regarding to the red LED in a high spectral efficiency system like KDPOF. Laser diode (red) causes an improvement of data-rate regarding to red LED due to higher injected power to fiber, higher cut-off frequency and narrower wavelength width. As drawback, LD has more dependency with temperature than LED, and LD is not eye-safe. These facts must be considered for the selection of the light emitter. On the other hand, a huge capacity gap can be observed between the current commercial products and KDPOF technology.

Long reach applications

For this case the system is configured to work with a symbol-rate of fS = 125 MHz. The transmitter uses a DAC of 12 bits and the receiver an ADC of 10 bits, both at symbol-rate. Next figure shows the performance as data-rate vs. standard SI-POF length for the different light emitters and launching NA.

Very long reach applications

In this case the system is configured to work with a symbol-rate of fS = 50 MHz. The transmitter uses a DAC of 12 bits and the receiver an ADC of 10 bits, both at symbol-rate.

Conclusions

From the presented performance results we conclude that cost-effective LED are enough to achieve rates of more than 1 Gbps over standard SI-POF in short reach applications. Furthermore, long and very long communication links can be implemented over standard SI-POF, delivering more than 400 Mbps at lengths around 200 meters and 100 Mbps at lengths around 300 meters.

On the other hand, capacity analysis and coupled design of TIA and digital front end show that green is the optimum color for SI-POF.

The performance results reported above could be modified in function of coupling of light-emitter with the fiber, POF manufacturer, bends, cuts and joins along the fiber, temperature conditions, etc. However, they show the potentiality of KDPOF technology and their advantages respect to current commercial products as high spectrally efficient and adaptive communication system for SI-POF.

[1] Ziemann, Krauser, Zamzow, Daum, “POF Handbook - Optical Short Range Transmissions Systems”, 2nd Edition. Springer

Ⓒ2010 Knowledge Development for POF S.A.