<March 2011> doc.: IEEE thz. Submission. Sebastian Priebe, TU Braunschweig

Transcription

1 Slide

2 Spatial and Temporal Dispersion in THz Indoor Propagation Channels Sebastian Priebe, Martin Jacob, Thomas Kürner Institut für Nachrichtentechnik, Technische Universität Braunschweig, Germany Slide 2

3 Outline. Introduction 2. Office Scenario 3. Spatial Dispersion 4. Temporal Dispersion 5. Summary/Outlook Slide 3

4 Introduction () Multipath propagation: has a high impact on THz indoor communication channels induces high spatial and temporal channel disperions is roughness-dependent, especially due to scattering Example of temporal dispersion: TX RX Relative received power [db] [ns] RMS delay spread relevant to estimate maximum symbol rates Slide 4 LOS path IEEE document thz_Towards_a_3_GHz_Channel_Model

5 Introduction (2) TX [ ] Example of spatial dispersion: RX [ ] S 2 [db] Spatial information relevant for MIMO communications Angular spread necessary for channel modeling H [db] LOS path only All paths f [GHz] IEEE document thz_Towards_a_3_GHz_Channel_Model Slide 5

6 Outline. Introduction 2. Office Scenario 3. Spatial Dispersion 4. Temporal Dispersion 5. Summary/Outlook Slide 6

7 Office Scenario Ray tracing simulations including scattering from rough plaster walls for two degrees of surface roughness plaster, less rough: l corr =.3 mm and σ h =.5 mm plaster 2, higher roughness: l corr =.7 mm and σ h =.5 mm 22 RX positions at equal distances of 25 cm and at a height of 75 cm LOS and NLOS conditions Omnidirectional antennas in vertical polarization!" Plaster walls /,-.!()*&+ #$%&&' 234#- (%&( Room height: 2.5 m,-. Slide 7

8 Outline. Introduction 2. Office Scenario 3. Spatial Dispersion 4. Temporal Dispersion 5. Summary/Outlook Slide 8

9 Spatial Dispersion () TX 2 Measure for the spatial dispersion provided by the angular spread: Only MPCs with a path loss up to 6 db considered in the following Angular spread σ ϕ,aoa,plaster [ ] σ ϕ = i 2 8 ( ϕ i ϕ ) 2 P i i P i TX σ ϕ,aoa,plaster 2 σ ϕ,aoa,plaster [ ] 2 5 y [m] 6 4 y [m] x [m] x [m] No significant roughness dependence of the angular spread Slide 9 5

10 Spatial Dispersion (2) TX/RX-distance-dependent angular spread model for plaster : σϕ,aoa[ ] LOS LOS Approximation NLOS NLOS Approximation /,-.!"!()*&+ #$%&&' 234#- (%&( d [m] AoA, azimuth,-. First values for d =.5 m due to height difference between TX and RX Good approximation with second order polynomial Prediction of spatial channel dispersion Slide

11 Spatial Dispersion (3) σθ,aoa[ ] LOS LOS Approximation NLOS NLOS Approximation σθ,aod[ ] LOS LOS Approximation NLOS NLOS Approximation d [m] d [m] AoA, elevation AoD, elevation Highest spreads close to walls, especially for the AoD Polynomial parameters including the second roughness of plaster 2 given in [] Slide

12 Spatial Dispersion (4) Omnidirectional antennas assumed so far But: highly directive (smart) antennas required for directed (N)LOS communications!" #" Selection of single multipath clusters Cluster-based modeling of angular spread obligatory Slide 2

13 Spatial Dispersion (5) Normalized histograms of the occuring cluster angular spreads evaluated individually for all clusters at every RX position: 2 Simulation Approximation Simulation Approximation PDF PDF σ ϕ,aoa [ ] σ ϕ,aoa [ ] Lower roughness Higher roughness Increasing spread for higher roughness Good approximation with a negative exponential distribution [] Slide 3

14 Spatial Dispersion (6) Comparison of approximated cumulative distribution functions: CDF Plaster Plaster 2 σ ϕ,aoa,plaster σ ϕ,aoa,plaster 2 σ θ,aoa,plaster σ θ,aoa,plaster 2 Almost identical behavior in azimuth and elevation AoD additionally analyzed in [] Randomization of angular cluster behavior for channel modeling [ ] Slide 4

15 Outline. Introduction 2. Office Scenario 3. Spatial Dispersion 4. Temporal Dispersion 5. Summary/Outlook Slide 5

16 Temporal Dispersion () Measure for the temporal dispersion given by the RMS delay spread: τ RMS = TX RMS delay spread [ns] 4 i ( τ i τ ) 2 P i i P i TX RMS, Plaster 2 RMS, Plaster [ns] y [m] 2 y [m] x [m] Highest RMS delay spreads under NLOS conditions Significant roughness dependence Slide x [m] 2.5

17 Temporal Dispersion (2) RMS [ns] Distance-dependent model for the RMS delay spread: d [m] LOS LOS Approximation NLOS NLOS Approximation RMS [ns] LOS LOS Approximation NLOS NLOS Approximation d [m] Lower roughness Higher roughness Several MPCs fall below the virtual noise level for the higher roughness Higher roughness leads to a lower RMS delay spread Slide 7

18 Temporal Dispersion (3) Maximum excess delay model: LOS LOS Approximation NLOS NLOS Approximation Max [ns] d [m] LOS LOS Approximation NLOS NLOS Approximation Max [ns] d [m] Lower roughness Higher roughness Maximum excess delay results from the longest multipath length Relevant longest MPC for the lower roughness drops below noise for the higher roughness Slide 8

19 Temporal Dispersion (4) Cluster-based modeling for directed (N)LOS communications Normalized histrogram of the cluster delay spreads for plaster :.45 Simulation Approximation.3!" PDF.5 #" [ps] RMS Good approximation with negative exponential distribution (parameters in []) Extremely low temporal dispersion achievable Slide 9

20 Temporal Dispersion (5) Comparison of the approximated CDFs for the two roughnesses:.8 CDF Plaster lower roughness Plaster 2 higher roughness [ps] RMS Far higher cluster-based RMS delay spread for higher roughness Slide 2

21 Temporal Dispersion (6) Estimation of max. symbol rates based on the RMS delay spread In approximation, no intersymbol interference (ISI), if: r s < K τ RMS Achievable symbol rates in the scenario: K: Constant between and ; here K = assumed for best case estimation CDF NLOS Plaster LOS NLOS LOS Plaster r S [GSymbols/s] No consideration of link budget aspects Better performance for higher roughness due to lower power of MPCs Higher symbol rates under LOS conditions Several Gbit/s only feasible with spatial filtering Slide 2

22 Temporal Dispersion (7) Evaluation of maximum symbol rates for each cluster in the scenario:.8 Plaster 2 (higher roughness) CDF.6.4 Plaster (lower roughness).2 Directed (N)LOS communications easily allow for several GSymbols/s regardless of the surface roughness At best, even up to, GSymbols/s can be achieved without intersymbol interference 2 r S [GSymbols/s] Slide 22

23 Outline. Introduction 2. Office Scenario 3. Spatial Dispersion 4. Temporal Dispersion 5. Summary/Outlook Slide 23

24 Summary/Outlook Ray tracing simulations have been performed in an office scenario An angular and an RMS delay spread model have been derived to predict the spatial and temporal THz channel dispersion High spatial dispersion occurs No influence of the two tables can be observed Single clusters allow for several GSymbols/s without intersymbol interference over directed communication links Next step: Development of a complete stochastic channel model including amplitude, phase and spatial as well as temporal information for system simulations Slide 24

25 References Parameters of the second order polynomials and the approximated analytical PDFs for both roughnesses can be found in: [] Priebe, S.; Jacob, M.; Kürner, T.: Angular and RMS Delay Spread Modeling in View of THz Indoor Communication Systems. 2th URSI Commission F Triennial Open Symposium on Radio Wave Propagation and Remote Sensing, 8 pages, Garmisch-Partenkirchen, March 2. Slide 25

26 Thank you for paying attention. Dipl.-Ing. Sebastian Priebe Slide 26