Digital Terrestrial TV Broadcast Current situation and future perspectives

Erik Stare Teracom Overview

1. About Teracom 2. Digital Terrestrial TV Broadcasting • Short tutorial • Standards 3. Challenges for terrestrial broadcasting 4. WiB – the basis for next-generation standards for terrestrial broadcasting (and broadband?)

2017-03-29 Erik Stare 2 What does Teracom do?

2017-03-29 Erik Stare 3 We use terrestrial transmitters…

2017-03-29 Erik Stare 4 … often at 300 m height…

2017-03-29 Erik Stare 5 …to allow for roof-top reception of digital TV…

… and in many cases in-door reception

We also transmit FM radio…

…and digital radio

2017-03-29 Erik Stare 6 We are located throughout

FM coverage TV coverage Backbone network

2017-03-29 Erik Stare 7 About Teracom

About the company (2017) Media & Broadcast Networks Public Pretection & Disaster Relief • Operates the digital terrestrial • Turnover: 1.987 GSEK • Operates about 6000 km of TV and radio networks in • Operates current TETRA-based • Result: 480 MSEK radiolink (microwave) Sweden and RAKEL network in Sweden • 560 employees • Uses about 6000 km of fibre in • Operates FM radio in Sweden • Responsible for the coastal • Founded in 1992 with old roots Sweden and Denmark radio/satellite system in Denmark from Televerket • About 900 sites • Media hub for 1000 TV • Own field service organisation • Co-location and data centers over the entire country channels

2018-04-28 8 Swedish DTT Network - Our customers

Pay-tv Operators

Broadcasters

2017-03-29 Erik Stare 9 Swedish DTT Network – Service mix

• 56 national program services: o 47 SDTV och 9 HDTV o 6 free-to air and 50 pay-tv services (Boxer)

2017-03-29 Erik Stare 10 2018-04-01 Swedish DTT network: services and multiplexes T2 T2 T2 Multiplex 1 22 Multiplex 2 20 Multiplex 3 20 Multiplex 5 33 Multiplex 6 37 Multiplex 7 31 06 reg (41) reg (35) NickJr 18 TV4 Film 2 reg (4) TV4 (12) TV12 (13) Fox reg (46) (61) SVT1 HD (48) (1/96) SVT1 (3) TV3 (8) TV8 reg reg (15) TNT 2 reg (19)National Geographic HD (37) (10) TV10 Cartoon Network (5) Kanal 5 (62) SVT2 HD

(25) Axess TV reg (2/97) SVT2 (14) MTV (27) BBC World News (55)C More Live HD/Hits HD (6) TV6 TV3 HD (26) CNN (63) 05 06 (54) (36) 21 reg Boomerang C More Fotboll/Hockey/Stars (34) 19 (7) (16) 06 (98) Barnkanalen/SVT24 Nickelodeon (39) 18 (37) Disney XD BBC Earth (64) TV4 HD (23) BBC Brit (9) (22) Investigation Discovery (21) Discovery Channel (51) (32) Travel Channel (11) Kanal 11 (99) (52) (65) Kanal 5 HD

(17) TLC (53) /SF-kanalen 06 (29) (24) TV4 Fakta (30) History Regional service 22 Disney (47) (38) VH1 (42) (20) Animal Planet HD Eurosport Channel (31) (28) Al Jazeera (82) fixed capacity Horse & Country TV

07 FTA (Free-to-air) Unscrambled service (NN) Logical Channel Number Time shared services MPEG2 SD based service MPEG4 HD based service Broadcast hours reg Regional insertions 23 Bergkvist/Tullstedt This document is the property of TERACOM AB and may not without our written permission be copied, imparted to a third party or used for any other unauthorized purpose. Swedish DTT Network - Infrastructure

• Mux* 1 (Public Service) 99,8% coverage** • Mux 2-7 98% coverage

• Main stations 54 (mux 1-7) • Intermediate stations 110 (mux 1-7) • Small sites 416 (mux 1)

* DTT Multiplex = A bundle of TV services that have been digitized, compressed and combined into a data-stream for transmission to the consumer over a single channel ** Population coverage 2017-03-29 Erik Stare 12 Swedish DTT Network – Customer offerings

DTT Broadcast: All customers Basic DTT Broadcast - 98% coverage - Basic SLA - MPEG2/SD, MPEG4/SD, MPEG4/HD

SVT Extended coverage Additional 1,8% coverage

SVT, TV4, TV3, Kanal 5 Regionalization Local content distribution (e.g. local + local services news and advertising)

SVT Enhanced SLA Improved SLA (reliability, security etc)

2017-03-29 Erik Stare 13 Swedish DTT Network – Regionalization 2017

Broadcast service Regionalization SVT1 / SVT2 21 + 21 regions

SVT1 HD / SVT2 HD 19 + 19 regions

TV3 19 regions

TV3 HD 19 regions

TV4 28 regions

TV4 HD 28 regions

Kanal 5 19 regions

Local TV services 3 services

2017-03-29 Erik Stare 14 Very short tutorial about digital terrestrial TV

2017-03-29 Erik Stare 15 Simplified transmission chain for digital terrestrial TV RF MPEG-2 TS Uncompressed (”air interface”) video Video coding Uncompressed audio Audio coding Modulation Mux

Other services

Channel

Decompressed video Video decoding De- Decompressed mux Demodulation audio Audio decoding

2017-03-29 Erik Stare 16 Constant and variable bit rate

• The efficiency of video coding / bit rate reduction depends on a number of factors, e.g. how ”difficult” the material is • Because the criticality of the material varies over time one gets the following relations: ‒ With constant bit rate (CBR) the quality varies over time ‒ With constant quality one gets a variable bit rate (VBR) ‒ None of these are desirable!

Picture quality, Q CBR Picture quality, Q VBR

tid tid

B, bit rate Q, Subjective picture quality 2017-03-29 Erik Stare 17 Statistical multiplexing • A large number of variable-bit rate (VBR) video services are combined into a stream that has both constant bit rate and a constant video quality

Capacity Mbit/s

Service Information, CA, bootloading etc TV service 4

TV service 3

TV service 2

TV service 1 Time 2017-03-29 Erik Stare 18 Multiplexing • The result of the audio/video coding is put in so-called MPEG-2 Transport Stream packets (TS packets) • The multiplexing operation assembles TS packets from different services to one single data stream – the MPEG-2 Transport Stream (MPEG-2 TS) ‒ One ”colour” per service component (e.g. ”video of SVT2” or ”audio of SVT1”) • This stream is broadcast over the air by the modulator/transmitter and is demultiplexed by the receiver

188 byte = 188 x 8 bits per TS packet

… TS packet 1 TS packet 2 TS packet 3 … MPEG-2 Transport Stream (TS)

2017-03-29 Erik Stare 19 Modulation RF signal MPEG-2 TS UHF: 8 MHz (ch.21-48) Constant bit rate VHF: 7 MHz (ch. 5-12)

Uncompressed video Video DVB-T or DVB-T2 coding Uncompressed Mux Modulation audio Audio coding

Channel

Decompressed video Video decoding De- Decompressed mux Demodulation audio Audio decoding

2017-03-29 Erik Stare 20 Spectrum for Digital Terrestrial TV

• UHF band IV/V channel 21-48: 470-694 MHz Status today of • VHF band III channel 5-12: 174-230 MHz 700 MHz band release (green=no DTT UHF band IV/V in 700 MHz band)

470 MHz 694 MHz 790 MHz 862 MHz

21 22 … … … … … 47 48 49 … 60 61 … 69

frequency

8 MHz 700 MHz 800 MHz Band band (fully released (earlier allocated VHF band III 31 Oct 2017) to 4G/LTE)

5 6 7 8 9 10 11 12

frequency 174 MHz 230 MHz

2017-03-29 Erik Stare 7 MHz 21 Worldwide Digital Terrestrial Broadcast Standards

2017-03-29 Erik Stare 22 DVB-T and DVB-T2 • DVB-T (1997) was the first emission standard for digital terrestrial TV ‒ 8 MHz RF bandwidth ‒ Possible to choose trade-off between capacity and C/N (C/I) performance ‒ Used in mux 1-5 ‒ Capacity = 22.1 Mbit/s with same coverage as analogue TV • DVB-T2 (2009) is based on DVB-T but with a lot of new functionality • DVB-T2 allows for about 50% higher capacity than DVB-T for the same coverage • After release of 700 MHz band (after 31 Oct 2017) ‒ 3 muxes with DVB-T (SD with MPEG-2/MPEG-4 video coding) ‒ 3 muxes with DVB-T2 (SD & HD with MPEG-4 video coding) ‒ Total broadcast capacity per site = 163 Mbit/s

2017-03-29 Erik Stare 23 DVB-T och DVB-T2 use OFDM

Representation of OFDM in the time-frequency plane

OFDM spectrum dB 10 Guard Interval (GI)/Cyclic prefix (CP) 0 used to protect against multipath -10

-20

-30 2 k mode -40

powerspectrum density -50 8 k mode

-60 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 MHz frequency relative to centre frequency f c 2017-03-29 Erik Stare 24 DVB-T2 builds on DVB-T

• OFDM based (thousends of orthogonal carriers) • Same basic OFDM parameters as DVB-T ‒ FFT size ‒ Guard interval ‒ Pilot patterns • But also many new values • Many other additions and improvements • A lot of the signal processing in the receiver is similar to DVB-T • T2 receivers also support DVB-T

2017-03-29 Erik Stare 25 Symbol time (FFT size) and guard interval

• With DVB-T2 the symbol time can be increased by a factor two (16K FFT) and four (32K FFT) compared to DVB-T ‒ Reduces the overhead due to guard interval for a given size of guard interval (size of SFN)  increased capacity

≈ 1 ms

GI 8K symbol 25% overhead in DVB-T with maximum guard interval

≈ 4 ms

~6% overhead GI 32K symbol in DVB-T2

• Increases possible guard interval size and therefore size of SFN for a given percentage GI overhead  potentially more efficient frequency plan • DVB-T2 may also use the same symbol periods as DVB-T (8K, 4K, 2K) 2017-03-29 Erik Stare 26 Flexibility in pilot pattern

• DVB-T has a fixed pattern of scattered pilot cells • DVB-T2 has 8 different patterns to choose from, depending on network type and reception conditions • Minimises pilot overhead

2017-03-29 Erik Stare 27 Forward Error Correction (FEC)

• DVB-T has a convolutional code + Reed-Solomon • DVB-T2 has an LDPC code + BCH code ‒ Same as in DVB-S2 (satellite) and DVB-C2 (cable) ‒ Iterative decoding of LDPC ‒ 6 code rates: 1/2, 3/5, 2/3, 3/4, 4/5, 5/6 ‒ Flexibility to make desired trade-off between capacity and robustness

‒ FEC block size (Nldpc): 64800 bits or 16200 bits

Nbch= Kldpc

Kbch Nbch-Kbch Nldpc-Kldpc

BBFRAME BCHFEC LDPCFEC

(Nldpc bits)

2017-03-29 Erik Stare 28 Interleaving • Interleaving is of fundamental importance for the RF performance on non-AWGN channels • DVB-T2 has several interleavers ‒ Bit interleaver within a FEC block ‒ Cell interleaver within a FEC block ‒ Time interleaving within ‒ Frequency interleaving within an OFDM symbol • The result is that bit errors caused by the channel are equally distributed among the FEC blocks, and also within FEC blocks  maximises error correction ability of the LDPC/BCH code

2017-03-29 Erik Stare 29 Interleaving in DVB-T and DVB-T2

Kmax = 1 704 if 2K Kmin=0 Kmax = 6 816 if 8K DVB-T FEC FEC ..... FEC ..... FEC ..... Single erased OFDM-symbol ......  Bit errors ...... FEC ......

TPS pilots and continual pilots between Kmin and Kmax are not indicated boosted pilot data

DVB-T2

Time Single erased OFDM-symbol FEC  Can be corrected!

Erik2017 Stare-03-29 30 Modulation

• T2 has a 256-QAM mode 256-QAM ‒ Carries 8 bits per data cell • The T2 standard also includes ‒ 64-QAM ‒ 16-QAM ‒ QPSK ‒ … inherited from DVB-T

2017-03-29 Erik Stare 31 Performance for DVB-T2 modulation and FEC close to theoretical limits

Capacity Performance

10,00

9,00

8,00

7,00 Capacity limits for a channel 6,00 with white noise (AWGN) 5,00

4,00

Effective Effective bits per Cell 3,00

2,00

1,00

0,00 0,0 5,0 10,0 15,0 20,0 25,0 30,0 C/N

DVB-T2 QPSK DVB-T2 16-QAM DVB-T2 64-QAM DVB-T2 256-QAM Shannon Limit BICM Limit

2017-03-29 Erik Stare 32 Challenges for the future

2017-03-29 Erik Stare 33 Some issues with current DTT • Reduced spectrum • Increased-quality services (SDHDUHD) • Cost reduction needed • Energy reduction needed • Local services vs Single Frequency Networks • Difficult to handle UHDTV efficiently • Network optimized for roof-top reception ‒ mobile reception requires dedicated optimised “mobile signals” ‒ or “half-good” compromise • Very difficult to introduce X-polar MIMO ‒ Due to existing H-only receiving antennas • No broadband (bi-directional point-to-point/unicast) • Relatively “isolated” from mobile telecom ‒ Not implemented in handheld devices

2017-03-29 Erik Stare 34 Migration to new standards

• Painful process to migrate to new broadcast standards • Difficult to justify a new “DVB-T3” standard without radically improved performance & functionality

• A small step is not enough… • Is a “giant leap” possible?

2017-03-29 Erik Stare 35 WiB – A new system concept for DTT

• Developed at Teracom • IBC2016 Best Conference Paper Award • May potentially resolve all identifies issues

2017-03-29 Erik Stare 36 How could that be achieved?

2017-03-29 Erik Stare 37 Traditional frequency planning

300 4

4 3 5 • Only a fraction of the UHF channels are used 3 5 200 1 1 2 6 2 6 7 4 100 from a given transmitter (TX) site 7 4 3 5 th 4 3 5 1 0 3 5 2 6 ‒ Typically about 1/5 of the channels (reuse-5) km 1 1 2 6 7 2 6 4 -100 7 7 4 3 5

3 5 -200 1 1 2 6 2 6 7

-300 7

-400 -300 -200 -100 0 100 200 300 400 km • In order to get sufficiently high total DTT capacity high order constellations are used ‒ DVB-T: 64-QAM ‒ DVB-T2: 256-QAM • These require a high SNR and therefore a lot of power

2017-03-29 Erik Stare 38 Required TX power for traditional DTT Extremely unbalanced RF power across UHF channels – Power [W] (drawn to scale) sub-optimum from efficiency point of view!

2500

DVB-T2 DVB-T2 DVB-T2 Mux 1 Mux 2 Mux 6

No power

Frequency

2017-03-29 UHF1Erik StareUHF2 UHF3 UHF4 UHF5 UHF6 UHF7 UHF8 … UHF24 UHF25 UHF26 UHF27 UHF28 39 Shannon’s law and required power

푹 = 푩 ∙ 풍풐품ퟐ(ퟏ + 푺푵푹) [bits/s] • Required power (SNR) increases exponentially with capacity (spectral efficiency) • High capacity also means high sensitivity to interference • Increased power not a good way to further increase overall capacity

2017-03-29 Erik Stare 40 What about the other way around?

• Using instead a more robust transmission mode (lower required SNR) ‒ Going backwards on the exponential curve! ‒ Allows for a large reduction in power • Allows using a lower reuse factor, i.e. more spectrum per site ‒ Larger basic signal bandwidth • Why not go all the way to QPSK and reuse-1?

Increased spectrum exploitation (reuse-1)

Reduced power 2017-03-29 Erik Stare 41 WiB - Spreading the power equally over all frequencies (Single wideband WiB signal from all sites, reuse-1) Power [W] (drawn to scale)

2500

DVB-T2 DVB-T2 DVB-T2 Mux 1 Mux 2 … Mux 6

17 dB difference per RF channel Factor 50!

WiB Frequency 50 … UHF1 UHF2 UHF3 UHF4 UHF5 UHF6 UHF7 UHF8 UHF9 UHF10 UHF24 UHF25 UHF26 UHF27 UHF28

2017-03-29 Erik Stare About 90% less total TX power by using all frequencies 42 Basic principles of WiB

Power [W] • Wideband (drawn to scale) ‒ Wideband transmission as a single WiB signal 2500 DVB-T2 DVB-T2 … DVB-T2 ‒ Covering potentially the whole 224 MHz UHF Mux 1 Mux 2 Mux 6 band (28 UHF channels) • Reuse-1 ‒ Adjacent TXs use the same frequencies ‒ Very challenging interference situation WiB Frequency ‒ e.g. C/I = 0 dB 50 … UHF1 UHF2 UHF3 UHF4 UHF5 UHF6 UHF7 UHF8 UHF9 UHF10 UHF24 UHF25 UHF26 UHF27 UHF28

• Robust transmission mode required Time ‒ e.g. QPSK, req. C/N close to 0 dB • Interference Cancellation ‒ Removes unwanted interference • Frequency hopping (optional) RF7 RF6 RF5 RF4 RF3 RF2 RF1

4*8=32 MHz Frequency

2017-03-29 Erik Stare 43 How to handle interference

• High basic robustness (close to C/I=0 dB) TX2 • Rejection via RX antenna ‒Rooftop: Directional antenna  Antenna discrimination 16 dB (ITU) RX ‒Mobile: Dynamic beamforming TX1

• Interference cancellation TX3

TX = Transmitter site RX = Receiver 2017-03-29 Erik Stare 44 Antenna beamforming Interference cancellation TX1 Via Successive Interference Cancellation and/or Receiver Antenna Beamforming/Cancellation RX Cancellation of TX2

TX2 TX2 All TXs are synchronised but Required C/N = 0 dB (linear 1) RX with different content and TX1 pilots TX3

Demodulated C1=4 and cancelled

Demodulated C2=2 C2=2 and cancelled

C3=1 C3=1 C3=1 Demodulated N=1 N=1 N=1

2017-03-29 Erik Stare Antenna beamforming may be independently optimised for each cancellation45step! Receiver complexity

• A receiver is not expected to demodulate the 224 MHz WiB signal (200-300 Mbps) as a whole ‒ A receiver rather extracts a selected service and demodulates only the associated part of the signal ‒ Only (e.g.) 4*8 MHz = 32 MHz at a time, then frequency hopping to new 32 MHz block etc ‒ Additional complexity for frequency-hopping tuner (e.g. TFS) is low

Time

RF7 RF6 RF5 RF4 RF3 RF2 RF1

4*8=32 MHz Frequency

• What we do have: ‒ Factor 4 increase in sampling frequency and FFT size due to wider tuner bandwidth ‒ Additional complexity for Interference Cancellation ‒ but rather limited thanks to all TXs being synchronized ‒ small loop rather than full remodulation

2017-03-29 Erik Stare 46 Network performance simulations

• Effective TX antenna height 250 m • 60 km TX separation • 1 kW ERP per UHF channel (17 dB lower than today) • Propagation according to ITU-R P.1546 • Standard deviation: 5.5 dB (shadow fading) + 2.0 dB (frequency-dependent fading) • Spatial correlation model • Three different time correlation models (C, U1, U2) • Directional RX antenna at 10 m (11 dBd gain, max 16 dB discrimination) • Best TX case: The best TX is chosen irrespective of content • Wanted TX case: A particular TX (with desired content) is required • Interference cancellation of up to 2 TX signals • Spectral efficiency calculated as average (normalized) Shannon capacity (95% probability, 99% of time) in the worst point

Time correlation type Best TX Wanted TX Inter/Intra site (C) 3.41 bps/Hz 1.55 bps/Hz DVB-T2 today: about 1 bps/Hz Intra-site (U1) 3.38 bps/Hz 1.37 bps/Hz No correlation (U2) 4.07 bps/Hz 1.60 bps/Hz 2017-03-29 Erik Stare 47 High bit rate services

• Thanks to the wideband nature of WiB, the system can efficiently handle video services with high peak data rate (28-40 Mbps) ‒ such as UHD • This includes also close-to-ideal handling of variable-bit rate services (statistical multiplexing)

2017-03-29 Erik Stare 48 Reduced costs

• Capital Expenditures (CAPEX) ‒ Single wideband transmitter (TX) ‒ Required total output power about half of one existing DTT TX ‒ No need for combiners - only a single wideband RF filter ‒ Lower equipment volume/weight Combiner room today ‒ May allow mast positioning of the TX  no RF feeder needed ‒ Lower performance requirements on TXs (linearity etc), due to robust transmission ‒ Drastically reduced need for cooling and backup power

• Operational Expenditures (OPEX) ‒ >90% reduction in fundamental energy consumption ‒ Reduced maintenance need (less equipment, less sensitive, longer lifetime) ‒ No need for frequency planning and frequency changes

Additional note: • Combiners and feeder together amount to about 3 dB attenuation • Getting rid of these allows for a further halving of power consumption  A total of 95% reduction !

2017-03-29 Erik Stare 49 Extension of the basic WiB concept (examples)

• Cross-polar MIMO (H + V polarisation on the same frequency) ‒ May further double the WiB capacity ‒ Could be backwards-compatible with legacy RX antennas ‒ Existing antennas would get ”normal WiB capacity” ‒ Sufficient separation via RX antenna polarization discrimination (16 dB) ‒ User with a new antenna could get twice the capacity

• Superposition-based combination of broadcast and unicast (mobile telecom) in the same spectrum ‒ Transmission on the same time/frequency (e.g. on the same ”resource block”) with controlled power difference ‒ Separated in the receiver by (LDM-based) successive interference cancellation

2017-03-29 Erik Stare 50 Non-Orthogonal Multiple Access (NOMA)

• FDM: Each signals may only use a subset of spectrum, but all time • TDM: Each signal may use all spectrum but only a subset of time (time slot) • NOMA (aka LDM): Each signal may use all spectrum all time ‒ Several signals are superimposed using the same time/frequency (e.g. resource block) ‒ One layer may be broadcast/multicast ‒ Several possible unicast layers

Demodulated C1=4 and cancelled

Demodulated C2=2 C2=2 and cancelled

C3=1 C3=1 C3=1 Demodulated N=1 N=1 N=1

2017-03-29 Erik Stare 51 Instead of this prolonged tug of war…

DTT spectrum Mobile Telecom spectrum

2017-03-29 Erik Stare 52 … why not this Win-Win peace project?

Controlled level DTT Mobile Telecom signals are distance ”invisible” for DTT receivers Separated via Mobile Telecom receivers first Interference Mobile Telecom demodulate and cancel DTT Cancellation

NOMA/LDM transmission using the same spectrum (100% of time, 100% of frequency) 2017-03-29 Erik Stare 53 A WiB Vision Same system/standard for broadcast and unicast

5G New Radio - Broadcast

5G New Radio - Unicast

Same NOMA/LDM-based system/standard

2017-03-29 Erik Stare 54 New advanced antennas with (e.g.)

DTT Horizontal polarisation • 18 horizontal 20-degree sectors (DVB-T/T2) • Vertical lobe width < 1 degree

Vertical polarisation WiB

Several superimposed Vertical Vertical Vertical Vertical signals within each sector Sector Sector Sector Sector #1 #2 #3 #4 using LDM

DTT vs WiB: • WiB unlikely to significantly affect current DTT • WiB @ 17 dB lower power/RF channel operation! • 16 dB polarisation discrimination • DTT may have some negative impact on WiB, but

2017-03-29 Erik Stare limited due to WiB wideband nature 55 C (10 km) B (20 km) A (30 km) WiB-BC + WiB-UC WiB-BC 3 dB (Example) WiB-UC

WiB-BC 20 dB 3 dB TX C B A Horizontal WiB-UC Sector (20°)

10 dB

WiB-BC 3 dB N WiB-UC

• WiB-UC may exploit the (far) better C/(N+I) closer to the TX • Users A, B and C may also use the same spectrum in LDM (more efficient than TDF/FDM)

2017-03-29 Erik Stare 56 Possible standardisation paths for WiB

DVB Commercial and Technical Study Missions about WiB (until end of May 2018)

3GPP-5G DVB Standard? Standard?

2017-03-29 Erik Stare 57 Big enough leap? WiB – Summary of gains • Increased spectral efficiency • Radically reduced network cost ‒ and energy consumption • Unconstrained use of local services • Allows for efficient handling of UHD ‒ Close-to-ideal video coding statmux gain (VBR services) • Doppler performance allowing high-speed mobile reception of all “roof- top” services • Broadcast and broadband in the same spectrum • Converged win-win solution with mobile telecom

2017-03-29 Erik Stare 58 IBC2016 Best Conference Paper Award to WiB

From left • Erik Stare, Teracom • Peter Klenner, Panasonic • Jordi Giménez, UPV

From award ceremony 11 Sept 2016 at IBC, Amsterdam 2017-03-29 Erik Stare 59 Thank you for your attention!

2017-03-29 Erik Stare 60