software-defined white-space cognitive systems: implementation of the spectrum sensing unit
DESCRIPTION
S. Benco, F. Crespi, A. Ghittino, A. Perotti, "Software-defined white-space cognitive systems: implementation of the spectrum sensing unit", Proceedings of the 2nd International Workshop of COST Action IC0902 October 5–7 2011, Castelldefels and Barcelona, SpainTRANSCRIPT
Software-defined white-space cognitive systems:
Implementation of the spectrum sensing unit
Castelldefels, October 6th 2011
Sergio Benco, Floriana Crespi, Andrea Ghittino, Alberto Perotti
Integrated Networks Laboratory (INLAB), CSP s.c.a r.l. - ICT innovation,TURIN (ITALY)
Software-defined white-space cognitive systems 2
Outline
TV White-Spaces Spectrum Sensing
IEEE 802.22 Spectrum Sensing model
DVB-T CP autocorrelation Spectrum Sensing
Threshold calculation
Performance
Conclusions and future work
Software-defined white-space cognitive systems 3
Spectrum sensing for TV White-Spaces
TV White-Spaces represent the area (space domain) and the portion of the spectrum (VHF and UHF bands) where the broadcast signal strength falls below the sensitivity level of Primary User (PU) receivers
Regulatory bodies are currently discussing about Secondary User (SU) spectrum sensing requirements in order to avoid interference to DVB-T receivers
Interference issues can be faced through:
SU geo-location and PU database queries Cognitive Pilot Channel (CPC) SU autonomous sensing (cooperative or not)
Software-defined white-space cognitive systems 4
IEEE 802.22 Spectrum Sensing model
PU (TX)
SU
SUKeep-out region (R
km)
Range at wich the Desired/Undesired (D/U) ratio falls below 23 dB
SU spectrum sensing requirements: ● PU Rx characteristics: F/B = 14 dB; D/U = 23 dB● At PU Rx: P
RSU ≤ P
RPU – D/U
dB + F/B
dB
● At PU Rx: PR
SU ≤ -101 dBm At SU Rx: Sens. ≤ -115 dBm
PU (RX)
PU protection contour (Dkm
)
Sensitivity range of the PU Rx ITU-R: P
RPU = -92dBm @ 132km
ERP TX = +90dBm, height: 500m, 615MHz)
A SU must detect a PU Tx at a range of: Rkm
+ Dkm
Software-defined white-space cognitive systems 5
DVB-T spectrum sensing: CP autocorrelation
N0CP
N0d
N1CP N1
d
Ns
DVB-T sensing module parameters
Modes 8k (6817 subcarriers)2k (1705 subcarriers)
CP lengths 1/4, 1/8, 1/16, 1/32
Channel bandwidth 8 MHz
Sampling rate 12.5 MS/s (12.5 MHz)
R xx(i) = ∑k=0
K−1
∑n=i+kN s
i+kN s+N cp−1
x (n) x (n+N d)CP correlator:SeeReferences (1)(2)
T CP=max
i∣Rxx(i)∣
Avgi∈J
∣Rxx(i)∣⩾<
γ
CP correlator test:
Ns Symbol samplesNCP CP samplesNd Data samplesK Number of symbols
Software-defined white-space cognitive systems 6
DVB-T spectrum sensing: applied threshold
γ = γ ⋅ Avgi∈J
∣R xx(i)∣ ⇒ PFA⩽ 0.1
T CP=max
i∣Rxx(i)∣
Avgi∈J
∣Rxx (i)∣= θ
∣Rxx∣⩾<
γ
J=N ∖QN={n∈ℕ : 0 ⩽ n < N s}
Q={q∈ℕ : θ−NCP ⩽ q < θ+N CP}
False Alarm Probability (P
FA )
obtained through Monte-Carlo simulations over 1000 trials
The threshold is adaptive w.r.t. the actual average correlation level plus a fixed margin that depends on P
FA
Software-defined white-space cognitive systems 7
DVB-T spectrum sensing over K symbols
Symbol synchronization permits to obtain a coherent combining and average over K subsequent DVB-T symbols thus achieving a processing gain of about 5 dB for each 10 dB increase in K
1 symbolSNR = -15dB
AWGN channel
10 symbolsSNR = -15dB
AWGN channel
100 symbolsSNR = -15dB
AWGN channel
Software-defined white-space cognitive systems 8
DVB-T OFDM sensing: performance
The detection time Tdet of this real-time module is calculated at the target sensing performance (PFA=0.1, PD=0.9) for a given SNR:
SNR (PD=0.9) = -17 dBSymbols = 100Tdet = 112.00 ms + Tproc
SNR (PD=0.9) = -12 dBSymbols = 10Tdet = 11.20 ms + Tproc
Tch move time = 2000 ms
Tsensing = Tch move time – 2Tdet
Software-defined white-space cognitive systems 9
Conclusions and future work
● The DVB-T spectrum sensing based on CP autocorrelation offers a good trade-off between complexity and effectiveness
● The first attempts to exploit TV white space have raised the problem of high sensitivity requirements for the SU spectrum sensing unit
● We have developed a real-time module for OFDM spectrum sensing that approaches the requirements for the IEEE 802.22 WRAN spectrum sensing unit
● Future work will provide a SU network able to continuously monitor the TV White-Spaces through a CP-based spectrum sensing module using the GNURadio/USRP2 platform
Software-defined white-space cognitive systems 10
References
(1) D. Danev, E. Axell, and E. G. Larsson, “Spectrum Sensing Methods for Detection of DVB-T Signals in AWGN and Fading Channels”, In Proc. IEEE PIMRC, pp. 2721-2726, Dec. 2010
(2) V. Gaddam, M. Ghosh, “Robust Sensing of DVB-T Signals”, New Frontiers in Dynamic Spectrum, 2010 IEEE Symposium on , vol., no., pp.1-8, 6-9 April 2010
(3) S. Shellhammer, V. Tawil, G. Chouinard, M. Muterspaugh, M. Ghosh, “Spectrum sensing simulation model”, IEEE 802.22-06/0028r10, Sept. 2006
(4) C.R. Stevenson, C. Cordeiro, E. Sofer, G. Chouinard, “Functional Requirements for the 802.22 WRAN Standard”, IEEE 802.22-05/0007r46, September 2005
Software-defined white-space cognitive systems 11
Contacts
Sergio Benco
Consulting Engineer, Integrated Networks Laboratory (INLAB)R&D dept.
mail: [email protected] cell: +39 329 0118356tel. +39 011-4815164
CSP innovation in ICT
Registered and Central Offices Environment Park - Laboratori A1via Livorno 60 - 10144 Torino
Operational OfficesVilla Gualino - Viale Settimio Severo 6310133 Torino
Tel +39 011 4815111Fax +39 011 4815001E-mail: [email protected]
www.csp.it