Simultaneous Data Transmission and Spectrum Sensing on Power Lines. Gautham Prasad and Lutz Lampe

Size: px

Start display at page:

Download "Simultaneous Data Transmission and Spectrum Sensing on Power Lines. Gautham Prasad and Lutz Lampe"

Silvia Fowler
5 years ago
Views:

1 Simultaneous Data Transmission and Spectrum Sensing on Power Lines Gautham Prasad and Lutz Lampe

2 Issue Electromagnetic Interference (EMI) Common mode (asymmetric) signals Unshielded wires 2-30 MHz: EMI with Broadcast radio - AM/DRM Amateur radio - HAM Other misc. - Citizens band, Aeronautical > 30 MHz: EMI with FM

3 Dynamic Spectrum Management Notches to accommodate non-plc services Dynamic notching in EN , ETSI TS Half-duplex (HD) operation requires listen-before-talk In-band full-duplex (IBFD) operation supports listen-and-talk Efficiency = t s T p HD: T p t s spent in spectrum sensing and waiting IBFD: t s = T p η = 1 1 see A. Schwager, Power Line Communications: Significant Technologies to Become Ready for Integration, Ph.D. dissertation, University of Duisburg-Essen, 2010

4 Spectrum Sensing Method Pre-defined bands for monitoring - based on services to accommodate PLC transmitter - secondary user in these bands Monitor noise power spectral density (PSD) on the line Use higher order FFT for better resolution - recommended: 9 khz RBW

5 Detection Conditions PSD across a 100 Ω load, dbm/hz (1) > 14 db (2) > -95 dbm Frequency, Hz x 10 7 The narrower spike corresponds to an AM signal The wider spike is a DRM signal In this case both spikes satisfy both criteria both interference signals are successfully detected See A. Schwager, Power Line Communications: Significant Technologies to Become Ready for Integration, Ph.D. dissertation, University of Duisburg-Essen, 2010

6 Operating under IBFD Self-interference caused by simultaneously transmitting while sensing the spectrum Monitor noise+selfinterference (PTX)dB PSD Transmitted signal ECG Level of received selfinterference is dependent on extent of echo cancellation gain (ECG) (PRSI+PN)dB (PN)dB IBFD monitoring signal HD monitoring signal P G. Prasad, L. Lampe, S. Shekhar, Enhancing transmission efficiency of broadband PLC systems with in-band full duplexing, IEEE ISPLC, 2016 G. Prasad, L. Lampe, S. Shekhar, In-band full duplexing broadband power line communications, IEEE Trans. on Commun., September 2016

IBFD Solution for BB-PLC X(l) Modulator LMS Weights G(l) E(l) Demodulator IDFT IDFT DFT DFT e(n)

)[n] + (x SOI *h PLC )[n] + w[n] RX-end AFE Passive cancellation - analog hybrid Active

Shekhar, Enhancing transmission efficiency of broadband PLC systems with in-band full duplexing,

7 IBFD Solution for BB-PLC X(l) Modulator LMS Weights G(l) E(l) Demodulator IDFT IDFT DFT DFT e(n) x _ + x SOI TX-end AFE P1 Power Line Channel (h PLC ) Active hybrid h SI P2 P3 y[n] = (x*h SI )[n] + (x SOI *h PLC )[n] + w[n] RX-end AFE Passive cancellation - analog hybrid Active cancellation - digital LMS filter G. Prasad, L. Lampe, S. Shekhar, Enhancing transmission efficiency of broadband PLC systems with in-band full duplexing, IEEE ISPLC, 2016 G. Prasad, L. Lampe, S. Shekhar, In-band full duplexing broadband power line communications, IEEE Trans. on Commun., September 2016

8 Echo Cancellation Gain Echo Cancellation Gain, db y[n] =(x h SI )[n] {z } echo Sub carrier attenuation, db +(x SOI h PLC )[n] + {z } s[n] {z} SOI cumulative noise ECG dependent on the strength of the received signal - i.e., PLC channel attenuation Higher ECG for higher attenuations No RX PLC signal in cognitive PLC operate under max ECG (~ 63 db)

9 Simulation Settings 2-30 MHz transmission bandwidth (P TX ) db = -50 dbm/hz (North American transmit mask) OFDM transmission FFT size = 3072 Sampling frequency = 75 MHz Spectrum sensing resolution bandwidth = 300 Hz Tested under a worst-case white noise of (P N ) db = -130 dbm/hz Under real PLC noise, the results are only better Interference signals generated as suggested by the test procedure in EN and ETSI TS

10 PSD Monitoring Results 85 PSD across a 100 Ω load, dbm/hz IBFD HD Frequency, Hz x 10 7 Increased noise floor -50 (TX) - 6 (hybrid) - 57 (digital cancellation) = -113 dbm/hz PSD across a 100 Ω load, dbm/hz IBFD HD Frequency, Hz x 10 7 Failed the >14 db condition with IBFD

11 Increased Noise Floor Increased noise floor in IBFD is a result of insufficient ECG Can only detect signals whose strength is above = -99 dbm/hz ECG limited by ADC dynamic range We propose three different solutions to counter this Solution (1): Use ADC of higher resolution

12 Impact of ADC Resolution 85 PSD across a 100 Ω load, dbm/hz bit 14 bit 16 bit Frequency, Hz x bit ADC sufficient to bring RSI down to about -130 dbm/hz Deployment requires new chipsets with higher ADC resolution

13 Exploiting Conductor Coupling Loss Higher passive isolation results in lower selfinterference power entering the ADC Lower self-interference strength produces smaller quantization noise (QN) at the ADC Solution (2): With sufficient passive isolation, a 12- bit ADC produces can bring the QN well below the noise floor P QN = P input SQNR

14 PSD Monitoring with better Isolation IBFD 2 wires HD PSD across a 100 Ω load, dbm/hz Frequency, Hz x 10 7 RSI brought down to the noise floor Solution uses more resources to accomplish SISO transmission

15 Analog Cancellation ECG limited by quantization noise introduced in the ADC P QN = P input SQNR Solution (3): Circumvent by cancelling the echo before ADC

16 PSD Monitoring with AC 85 PSD across a 100 Ω load, dbm/hz IBFD DC IBFD AC HD Frequency, Hz x 10 7 IBFD noise floor brought down to HD noise floor PSD across a 100 Ω load, dbm/hz IBFD DC IBFD AC HD Frequency, Hz x 10 7 Condition failed with DC, but passed with AC

17 Impact on Data Rates f end,m Z 1+ P TX(f) H PLC (f) 2 C = X m2 f start,m log 2 P RSI (f)+p N (f) df Φ is the set of the pre-defined AM/DRM bands With a transmit PSD of -50 dbm/hz and a constant low noise PSD of -130 dbm/hz, C = 77 Mbps * for a 40 db attenuation line with AC *a bit unrealistic but to give a number

18 Summary Overcame the drawbacks of using the state-of-the-art IBFD implementation for cognitive PLC Proposed three different solutions to counter insufficient ECG 1. Increase ADC resolution 2. Sense spectrum on other idle line(s) 3. Use analog echo cancellation Achieved 100% efficiency in cognitive PLC operations by eliminating sensing-only, waiting, and other idle times

Full-Duplex Spectrum Sensing in Broadband Power Line Communications

Full-Duplex Spectrum Sensing in Broadband Power Line Communications Gautham Prasad and Lutz Lampe Department of Electrical and Computer Engineering The University of British Columbia, Vancouver, BC Email: