[LINK] ABC Politics - Was : NextG Wireless Stays Up As NBN Fibre Broadband Crashes

[Tom Koltai]

 > -----Original Message-----
 > From: Tom Worthington [mailto:tom.worthington at tomw.net.au]
 > Sent: Monday, 7 February 2011 10:13 AM
 > To: Tom Koltai
 > Cc: link at anu.edu.au
 > Subject: Re: [LINK] FW: NextG Wireless Stays Up As NBN Fibre 
 > Broadband Crashes
 > 
 > 
 > Tom Koltai wrote Friday, 4 February 2011 12:37 PM
 > 
 > > The artefacts are possibly acoustic waves reaching the ionosphere 
 > > interfering with the uplink/downlink ...
 > 
 > No. The ABC TV picture showed a reporter in the foreground
 > and trees out 
 > a window behind them. When the trees waved in the wind, the picture 
 > broke up. When the trees stopped waving, the picture was 
 > okay. The most 
 > plausible reason for this was that when the trees waved, the image 
 > compression algorithm had more data than it could cope with. The 
 > solution I suggested to the ABC was: don't point the camera 
 > at the window.
 > 
 
 Any idiot can point a High Definition (HD) Video Camera at a 
 subject and pipe the output via WiFi to a router which is 
 then connected via 3G. 
 
 Some questions we should ask:
 
 What else in the room apart from the Sony Betacam was using 
 the same spectrum WiFi ?
 What are the 3G interconnect speeds of the closest Cell tower? 
 Are the Cell Towers backhaul links, Microwave, PRI's, FDDI 
 (don't laugh), Sonnet or Fibre?
 If PRI's, (How many E1's are used for backhaul)? Are the E1's bonded ?
 3G, Therefore, are in fact the backhaul interconnects T1's ? 
 Are the T1's bonded ?
 Are there any other spectrum issues in that region ? E.G. 
 1900 Mhz crosstalk ?
 Unicast or Multicast ?
 
 Did the ABC research any of these issues prior to allegedly 
 using Broadband or did they just "wing-it"?
  
 OK, now that we understand some of the questions we can 
 address your response to my humourous and cynical analysis of 
 the "ABC Cyclone Pixelation" affair.
 
 For normal outside Broadcast operations, Mpeg-2 VBR -v- 
 Mpeg-1 CBR used to be a choice that the man on the spot 
 (local editor - sometimes the soundman) determined depending 
 on whether there was a lot of movement or not.
 The streaming speed of the two is also a consideration when 
 deciding which standard to use. If for example, I attempted 
 to send HD quality over a transponder that was only set-up in 
 the eighties to handle VCD quality Mpeg-1 then I would expect 
 break-up in the signal.
 
 Televised Cricket for example is something that we would all 
 be familiar with and remember the perfect Mpeg-1 picture 
 until the camera followed the ball rapidly at which time 
 pixelation set in. But not on the Grass (which was fairly 
 static) but the fast moving faces of the crowd in the stands.
 
 > > However, I also believe that pixelation is a "special effect"
 > > utilised by some mixers to produce the "affect" of "We are in 
 > > a Cyclone affected area and here is the pixelation to prove 
 > > it" mindset.  ...
 > 
 > No, this looked like the real thing to me. I doubt that any 
 > TV channel would deliberately have the image of their 
 > reporter's face obscured.
 > 
 
 As I stipulated above, pixelation is never on the static 
 portion of the image, it is usually on the rapidly moving part.
 Had the pixelation in the ABC footage been on the trees 
 (which is where Mpeg-1 CBR pixelation occurs) then the 
 viewers wouldn't have known there was a cyclone with wildly 
 flapping trees.
 
 > > ... It could be that the ABC budget is VSAT uplink only...
 > 
 > More likely the ABC was doing what it said they were
 > doing: using a broadband link for sending the report.
 
 No they didn't. (I was making a joke about the VSAT) They may 
 have sent it via broadband, but before you respond, lets 
 discuss that for a moment.
 TCP-IP is packet based Network. It sends a packet, and waits 
 for the ACK from the destination. If it receives an ACK, it 
 sends the next packet. If it doesn't receive an ACK it 
 resends the same packet continuously until the mandated 
 time-out or it receives an ACK.
 
 In other words, any format sent via TCP-IP and spooled at the 
 other end (or streamed multi-cast live) would arrive with no 
 artefacts (especially not pixelation of the static portion of 
 the image) without technical incompetency or artificial 
 injection of CNN Newsworthy editing tricks.
 
 Now lets see if we can find any way to pixelate a Mpeg-2 
 image over technically sound and expertly set-up broadband 
 (which I expect from an organisation like the ABC).
 
 Yep, if we took an HD ....TS file and streamed it down a 
 shaped connection of only 1.5 Mb then we would loose at lest 
 25% of the content. (Mpeg-2 is at least 1800 Kbps - whereas 
 Mpeg-1 is up to 1800 Kbps.) Any idiot knows that you cant 
 squeeze Mpeg-2 into a T1 and it only just squeezes into an E1 
 - at the bottom end of the scale - i.e.: No fast panning 
 scenes or you may get to 10.08 Mbit/s which obviously - 
 sustained - won't fit in any Broadband I know of in Australia 
 (outside of a Telstra PR lab).
 
 So Tom, are you suggesting that the ABC is full of 
 incompetent Technicians that don't realise that you can't 
 live stream HD via a 1.5 Mb (T1) Cell tower interconnect?
 
 Because if you are, that in itself is almost as bad an indictment.
 
 Live interview segments need special tricks like PiP or 
 multiple screens to forego the "we are really in a cyclone" artefacts.
 
 E.g.: green screen at Standard definition with background 
 window shots spooled a few seconds earlier.  
 Or, overlay streaming of the Mpeg-1 interviewer on a green 
 screen against the Mpeg-1 of the wall with the PiP of the 
 Mpeg-2 Window picture of the Trees. 
 (I can do this on Sony Vegas on a single core XP machine.)  
 Or splitting the HD into four separate streams and 
 multicasting using different IP numbers on separate carriers.
 
 Tom we can ping pong this backwards and forwards till the 
 cows come home, if you insist, but as I was doing this back 
 in 1994 for a commercial TV station who chose Ausnet Services 
 to deliver BBC programming from the UK as an alternative to 
 Satellite - I have the anecdotes and empirical data to back 
 up my rather cynical view of the ABC's political Cyclone stunt.
 
 The reality is, Our current Broadband networks can't handle 
 HD Television.  We need the NBN. That was the impression that 
 broadcast left on everyone's hippocampus.
 
 Result:
 Actual real Pixelation - MEH Fail
 PR for the ongoing roll-out of the NBN - Pass, Distinction.
 
 Therefore, on the question in every tech persons minds... 
 Does the ABC employ idiots? 
 Nope.
 But they do appear to have some right winged psychologists 
 influencing their programming and content delivery methods (;-/).
 
 OR
 --
 They are low on Budget funds and are using broadband so as 
 not have to pay transponder fees. (On B3 that's $8,000 per 64 
 Kbit/s per month [eighteen cents per minute] - approximate 
 commercial rate [@ 10 Mbit/s = $ 228.05 per minute])
 
 
 References:
 MBONE: Multicasting Tomorrow's Internet http://www.savetz.com/mbone/
 A book about the multicasting backbone and the future of 
 multimedia on the Internet
 Copyright 1996, 1998 by Kevin Savetz, Neil Randall, and Yves Lepage
 
 Simplifying Headend Architectures
 http://www.motorola.com/web/Business/Solutions/Industry 
 Solutions/Service Providers/Telcos/Digital Headend 
 Solutions/MPEG-4 Encodiing for 
 Telcos/_Documents/staticfile/SP_SimplifyingHeadendArchitectures.pdf
 Motorolla White Paper 2009