1. Jayasri Akella Error Analysis of Multi-Hop Free-Space Optical Communication Jayasri Akella, Murat Yuksel, Shiv Kalyanaraman Department of Electrical, Computer and Systems Engineering Rensselaer Polytechnic Institute Troy, New York 12180 Email: akellj@rpi.edu

2. Jayasri Akella Motivation  To improve quality of Free Space Optical link  Communication medium being open space the link suffers from the vagaries of atmosphere impairing the link SNR, causing high end-to- end BER and high error variance.  Multihop approach reduces both end-to-end error and its variance and enables the design of efficient FEC schemes to improve the link reliability.

3. Jayasri Akella Outline  Introduction to FSO communication system  Effect of atmospheric on a single hop FSO link  Effect of atmosphere on multihop FSO link  Comparison  Conclusions

4. Jayasri Akella Free space optical communication - A brief introduction Line-of-sight communication technology using optical range (IR- Blue) of the EM spectrum Medium of transmission is free space/air.

5. Jayasri Akella Pros and Cons of FSO Communication  Pros  Easy to deploy in terms of cost and time  Very high bandwidth  Low power per bit  Cons  Should always maintain line of sight  Adverse atmospheric effects

6. Jayasri Akella Channel Behavior  FSO channel behaves like a time varying attenuator.  Causes of attenuation  Fixed geometric spreading  Atmospheric attenuation Fog: Can cause up to 300dB/KM Rain/Snow/Hail: Can cause up to 6db/KM (much less!!)  Causes of noise: ― Scintillation: Due to pockets of varying refractive index in atmosphere. ― Ambient light and thermal noise

7. Jayasri Akella Link Power Budget P Total Transmitted Optical Power at the transmitter P Rcvd Received Optical Power P Lens Losses at the lenses on both ends of the communication P Geometric Spread due to the finite divergence of the light beam P Atmospheric Attenuation caused by the suspended particles in atmosphere.

8. Jayasri Akella Effect of Atmosphere on FSO link  Rain/Snow  Fog  Size of optical wavelength is comparable to the size of fog particles. So the maximum attenuation experienced for fog ~300DB/Km (in contrast to RF, where rain causes the maximum damage to the signal.) sometimes leading to total loss.  Turbulence and Scintillation are the sources of noise. Effects of Rain/Snow and Fog can be can be captured in “Visibility”.

9. Jayasri Akella Effect of Atmosphere on FSO link

10. Jayasri Akella Effect of Atmosphere on FSO link

11. Jayasri Akella Error Probability due to Attenuation  For each packet, can model channel as constant since FSO channel is slowly varying.  For On-OFF keying the error probability is given by:  Where a v is atmospheric attenuation of channel

12. Jayasri Akella Error Probability over Single Hop

13. Jayasri Akella Visibility versus Number Hops

14. Jayasri Akella Reliability of the FSO link  To increase the reliability of an FSO link, two important methods have been proposed in the literature  Hybrid Approach: Provide hybrid link protection using an RF link [1]  Multi-hop approach: Scaling the hop length down between the transmitter and receiver using multi-hop routing[2].

15. Jayasri Akella Multihop Increases Efficiency of FEC schemes  FEC (forward error correction codes) can be used on top of multi-hop approach to improve link reliability.  If we manage to tightly bound error variance within certain limits, we can design more efficient error control codes for a given FSO link.  We show through simulations that multi-hop end-to-end error is lower and also has a smaller variance than single hop.

16. Jayasri Akella Channel Model For small errors P e <10e-2, the channel is approximated as:

17. Jayasri Akella Error Accumulation with Hop Length

18. Jayasri Akella Bit Error Rate versus Number of Hops Assume fixed link range

19. Jayasri Akella Transmitted Power versus Hop Length.

20. Jayasri Akella Simulation Details for Multi-hop scenario  Clear weather conditions :  Visibility is modeled as a Gaussian N~(10,3) Kms and variance 3 Kms (rough approximation from Albany, NY visibility data from the past 30 years.)  Adverse weather conditions:  Visibility is modeled as a Gaussian with mean 3 Kms and variance 1.5 Kms (rough approximation from Albany, NY visibility data from the past 30 years.)  Hop Length is 500 meters for multi-hop scenario, end-to-end range is 2.5 Kms (5 hops)

21. Jayasri Akella Single Hop and Multi-hop Error comparison Clear Weather Conditions

22. Jayasri Akella Single Hop and Multi-hop Error comparison Adverse Weather Conditions

23. Jayasri Akella Comparison Number of Hops Mean error Clear Weather Mean error Adverse Weather Variance Clear Weather Variance Adverse Weather 11.5e-30.270.020.1176 59e-275e-38e-504.5e-3 Multi-hop significantly outperforms single hop

24. Jayasri Akella Conclusions  The mean error is smaller over multiple hops compared to single hop for the same link range.  The variance is also smaller for the multi-hop case. Small variance helps to design efficient FEC schemes  Future Work:  Design suitable FEC schemes over multi-hop FSO link  Optimization of cost versus reliability for multiple hops