Presentation - Effective Transmission Schemes for Bandwidth Limited - PDF document

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/320427143 Presentation - Effective Transmission Schemes for Bandwidth Limited Satellite Operations Presentation · September 2017 DOI: 10.13140/RG.2.2.17294.66885 CITATIONS READS 0 16 4 authors: Manfred Ehresmann Florian Grabi Universität Stuttgart 15 PUBLICATIONS 25 CITATIONS 55 PUBLICATIONS 72 CITATIONS SEE PROFILE SEE PROFILE Georg Herdrich Rene Laufer Universität Stuttgart Luleå University of Technology 496 PUBLICATIONS 2,460 CITATIONS 117 PUBLICATIONS 234 CITATIONS SEE PROFILE SEE PROFILE Some of the authors of this publication are also working on these related projects: Developing new dust trajectory sensor with single channel amplifier View project Hybrid Ablative Development for Re-Entry in Planetary Atmospheric Thermal Protection (HYDRA) View project All content following this page was uploaded by Manfred Ehresmann on 16 October 2017. The user has requested enhancement of the downloaded file.

www.uni-stuttgart.de IAC-17-IAC-17,B4,3,6,x39641 Effective Transmission Schemes for Bandwidth Limited Satellite Operations M. Ehresmann, Florian Grabi, Georg Herdrich, René Laufer 68th International Astronautical Congress Adelaide, Australia 26.08.2017 26/8/2017 1

Effective Transmission Schemes for Bandwidth Limited Satellite Operations www.uni-stuttgart.de Introduction and technical context Part 1: Block data transmission • Bisection scheme • Gradient selective scheme • Performance evaluation Part 2: Continuous data transmission • Predictive corridor scheme • Performance evaluation Summary and Conclusions 68th International Astronautical Congress 26/8/2017 2 Adelaide, Australia

Introduction and technical context: MIRKA2 / CAPE www.uni-stuttgart.de MIRKA2-RX Transmission tests after flight on REXUS19 sounding rocket CAPE – CubeSat Atmospheric probe for education Limited power for transmission Limited visibility for ground stations MIRKA2-RX Capsule data rate MIRKA2 – Micro re-entry capsule 2 Maximum data rate 42.5 b/s Uncertain window for transmission Average data rate 12.1 b/s Limited data rate (Iridum – SDB) Minimum data rate 4.8 b/s 68th International Astronautical Congress 26/8/2017 3 Adelaide, Australia

Part 1: Block data transmission www.uni-stuttgart.de Transmitting a data set is a common case for satellite operations • Limited access to a ground station • Limited transmission capability of satellite (power) • Constraints on antenna pointing (attitude) Conventional case:  Chronological transmission of data Data set is transmitted by a First In First Out (FIFO) scheme or similar. In case of link break data is lost Potential reasons: Atmospheric conditions, bad pointing, power system overstrain  Reliability problem 68th International Astronautical Congress 26/8/2017 4 Adelaide, Australia

Part 1: Block data transmission – Bisection scheme www.uni-stuttgart.de Solution 1: Bisection scheme Fig. 1. Bisection scheme for an arbitrary signal and corresponding set of data points. Numbers indicate order of transmission of the measured data points. Change of order of transmitted data points Successive bisecting of the data set  Does converge to the full set  Global view is obtained quickly, resolution increases with time  Is deterministic: Can be hardcoded 68th International Astronautical Congress 26/8/2017 5 Adelaide, Australia

Part 1: Block data transmission – Bisection scheme www.uni-stuttgart.de Solution 2: Gradient selective scheme Figure 2: Gradient selective scheme for an arbitrary signal and corresponding set of data points. Numbers indicate order of transmission of the measured data points. Change of order of transmitted data points by local gradient: 𝑕 𝑗 = 𝑒𝑧 𝑗+1 𝑒𝑦 𝑗+1 − 𝑒𝑧 𝑗−1 𝑒𝑦 𝑗−1 = 𝑧 𝑗+1 − 𝑧 𝑗 𝑦 𝑗+1 − 𝑦 𝑗 − 𝑧 𝑗 − 𝑧 𝑗−1 𝑦 𝑗 − 𝑦 𝑗−1 .  Does converge to the full set  Global view of extremes is obtained quickly, resolution increases with time  Not deterministic: Index/time stamp of transmitted data point relevant 68th International Astronautical Congress 26/8/2017 6 Adelaide, Australia

Part 1: Block data transmission – Performance Evaluation www.uni-stuttgart.de Case: Random linear concatenated functions FIFO: Bisection: Gradient selective: • • Very quick global view Significant loss of data Quick global view • • when link is prematurely Reasonable Full convergence at terminated convergence at 25% 25% • Not always (!) Good convergence at 50 % 68th International Astronautical Congress 26/8/2017 7 Adelaide, Australia

Part 1: Block data transmission – Performance Evaluation www.uni-stuttgart.de Case: Random quadratic concatenated functions FIFO: Bisection: Gradient selective: • • Very quick global Significant loss of data Quick global view • when link is prematurely Reasonable extrema view • terminated convergence at 25% Good convergence at • 25% Good convergence at 50 % 68th International Astronautical Congress 26/8/2017 8 Adelaide, Australia

Part 1: Block data transmission – Performance Evaluation www.uni-stuttgart.de Other : Concatenated exponential functions: - Not well handled by bisection scheme, due to indifference to extrema Interesting parts might be skipped. - Well handled by gradient selective schemes, extrema covered first Concatenated sine functions – similar to noisy signals: - Well handled by bisection scheme, due to indifference to signal shape. - Not well handled by gradient selective scheme, too many sharp gradient changes.  Signal smoothing recommended. 68th International Astronautical Congress 26/8/2017 9 Adelaide, Australia

Part 2: Continuous data transmission www.uni-stuttgart.de For most satellite applications the overall transmission data rate is limited. The potential output data rate of all on-board sensors is much greater. Conventional solution: Reduce overall or sensor specific data rates to accommodate the available bandwidth • Potential undersampling of some sensors • Potential missing of some effects • Potential collection and transmission of uninteresting/irrelevant data “Uninteresting”  Monotonous / similar data to already known data. 68th International Astronautical Congress 26/8/2017 10 Adelaide, Australia

Part 2: Continuous data transmission www.uni-stuttgart.de Solution: Corridor predictive scheme • Minimization of individual sensor data rate • Selection for interesting data points • Calculate data prediction corridor  Data inside the corridor is “uninteresting” Coding scheme is known for ground operator  Implicit knowledge about not-transmitted data  Data outside the corridor is “interesting” and is transmitted  Follow-up recalculation of prediction corridor • Calculation scheme needs to be light-weight for OBC 68th International Astronautical Congress 26/8/2017 11 Adelaide, Australia

Part 2: Corridor predictive scheme – Linear regression www.uni-stuttgart.de Simple example: Linear corridors. Corridor boundaries: 𝑧 𝐷𝑝𝑠𝑠,𝑉𝑞/𝑀𝑝𝑥 = 𝑏 𝑦 + 𝑧 0 ± 𝑐 𝐷𝑝𝑠𝑠 Calculate linear regression function from initial data set. Losses defined by corridor width 𝑐 𝐷𝑝𝑠𝑠 Slope a: Violation condition: 𝑜 𝑇𝑏𝑛𝑞𝑚𝑓 (𝑦 𝑗 − Ӗ 𝑦) 𝑧 𝑗 − ധ 𝑧 σ 𝑗=1 𝑦 > 𝑧 𝐷𝑝𝑠𝑠,𝑉𝑞 𝑦 ˅ 𝑦 < 𝑧 𝐷𝑝𝑠𝑠,𝑀𝑝𝑥 𝑦 𝑏 = 𝑜 𝑇𝑏𝑛𝑞𝑚𝑓 (𝑦 𝑗 − Ӗ 𝑦)² σ 𝑗=1 Add robustness with trigger threshold: 𝑜 𝐷𝑝𝑠𝑠𝑗𝑒𝑝𝑠𝑊𝑗𝑝𝑚𝑏𝑢𝑗𝑝𝑡 > 𝑜 𝑈𝑠𝑗𝑕𝑕𝑓𝑠𝑈ℎ𝑠𝑓𝑡ℎ𝑝𝑚𝑒 y-Intersection: 𝑧 0 = ധ 𝑧 − 𝑏ന 𝑦 Add reliability by defining a maximum interval between selected data points: Spine of the prediction corridor: 𝑒 𝑛𝑏𝑦 𝑧 𝑇𝑞𝑗𝑜𝑓 = 𝑏 𝑦 + 𝑧 0  Important to determine whether a sensor is broken or nominal “uninteresting” 68th International Astronautical Congress 26/8/2017 12 Adelaide, Australia

Part 2: Corridor predictive scheme – Performance www.uni-stuttgart.de • Red points are selected for transmission. • Sample size 20 • Selection and recalculation after 2 nd corridor violation • Max distance between points is 30 300 data points polled 19 selected for transmission  Reduction rate: 15.8 ! Figure 14: Concatenated randomized noisy linear functions input signal with predictive corridor scheme at low sampling rate. 68th International Astronautical Congress 26/8/2017 13 Adelaide, Australia

Part 2: Corridor predictive scheme – Performance www.uni-stuttgart.de • Red points are selected for transmission. • Sample size 20 • Selection and recalculation after 2 nd corridor violation • Max distance between points is 30 3160 data points polled 235 selected for transmission  Reduction rate: 13.4 Higher reduction possible, by increasing gap distance. Figure 15: Concatenated randomized noisy linear functions input signal with predictive corridor scheme at high sampling rate. 68th International Astronautical Congress 26/8/2017 14 Adelaide, Australia