Wednesday, December 02, 2015

Part II Principles of Communications, 2015, to Dec 2

Wrapped up Traffic Management, including
idea of signaling protocols, and
space/time scales of different techniques;
and review of the course contents.

Planning to do revision classes next term.
Supervision notes will be linked to course web page after end of term.

[systems structures lecture moved to background]

some notes/problems from supervisions added
now

Monday, November 23, 2015

Part II Principles of Communications, 2015, to Nov 27

This week is

  • Shared Media
  • Mesh Network Capacity
  • Coding and Multihop Radio (Cope)

Two useful network coding primer and coded storage articles...
and just what is a shim?

Friday, November 20, 2015

Part II Principles of Communications, 2015, to Nov 20

this week, covered scheduling, work conservation, max/min fairness, and PGPS approximations + queue jump!

Monday, November 09, 2015

Part II Principles of Communications, 2015, to Nov 13

Optimization (Nov 9) - background paper:-
from Princeton group

TCP in the wild - Optical Carrier Speed
table + Reminder about R-squared

Schedulers&Rounds....up to WFQ.

Wednesday, November 04, 2015

Part II Principles of Communications, 2015, to Nov 6

Flow Control, Congestion Control,
open and closed loop systems

Control Theory -
time domain model
transform to frequency domain
recipies for laplace transforms
stability & efficiency
proportional, integral and differential (and PID) controllers.

Online slides (PPT and PDF) are ok -
apologies for symbol/font problem in some of the printed pages!

Friday, October 30, 2015

Part II Principles of Communications, 2015, to Oct 30

Previously, Compact routing, Central routing (fibbing to link state),&path vector

This week covered
interdomain routing, including AS paths and deadlocks,
for which see interdomain revision notes/slides
+
multicast,
+
random routing

Just started on error and flow control.
[As usual, wikipedia is great on gaussian elimination ]

Pretty much as per schedule

Friday, October 23, 2015

Part II Principles of Communications, 2015, to Oct 23

This week has been about inter-domain routing, and BGP.
Covered background business relationships (peering, customer/provider etc), the protocol, the attributes/selection, SPP, and discussed scaling, convergence, and problems (like bad gadgets, wedgies) - as always, data is out of date, but Geoff Huston to the rescue...

Thursday, October 15, 2015

Part II Principles of Communications, 2015

This week (to Dec 16) I should have covered the graph material (random graphs + alpha&beta models of small world/clustered graphs), and
Compact and Centralised routing - viz
http://www.cl.cam.ac.uk/teaching/1516/PrincComm/slides/schedule-2015.html

On graphs and networks there's a new really nice book by Jon Kleinberg


There's some lack of precision about the terms "small world" but basically, (wikipedia is your friend) a scale-free network refers to the power law degree distribution, and consequential small diameter (and some clustering), which leads to the small world property. Not all small world networks are scale free, but scale free networks are small world...

Sunday, June 28, 2015

The Smell of Memory

scientists in cambridge have recently figured out how to code the smell of memory, not just a smell of one thing that evokes a particular memory, but the underlying phenomenon that is memory itself. It turns out that it is not so abstract, and that there are only really 7 key parameters (somewhat like taste, which is, of course, related to smell in any case). Working backwards from the examples of smells that evoke memories, using a statistical technique called PCA, scientists can now code the entire space that is memory - so not only evoking a particular memory, but replaying everything at once.

Of course, there are severe dangers of synaesthesia with this technique, so it will only be allowed in key, socially beneficial situations, such as court cases, or TV interviews with politicians.

And it must be noted that there will inevitably be people who have an innate fear of smell - loosely related to the smell of fear, but far more disabling in this context, except where you'd like the right to be forgotten - de-oderant amnesia.

Thursday, February 26, 2015

Re-use ill-considered (twitter and hypothermia)

I've been reading a lot about ethics recently for a forthcoming workshop in Oxford on the topic.
It seems that a lot of the book keeping elements of ethics have been defined (I guess, perhaps, in reaction to the lamentably long list of unethical things that are done by all and sundry in online media commercial appropriation of the essence of you, and dodgy medical research re-owning folk knowledge on drugs from rainforest eco-systems, or testing stuff on people who have no political or economic choices...

So there's a long list of stuff we (as researchers) have to go through.....

but it seems to me that quite near the top of the list ought to be something about re-use

In computing (and in general in engineering), re-use of stuff (re-cycling harware, and re-cycling good ideas, or software) is regarded as a good thing - rep-purposing too - the internet came out of dual-use (re-purposing a survivable defense network, as a cheap extnesive global communications utility for the public)...

but the real elephant in the room for me is re-use of stuff in a new context without ethical consideration.

Let me give two examples (from the title of this post)

1/ someone (NOT ill-advisedly) tweets that, unless the snow in an airport is cleared soon, he will blow it up. This is a daft joke which maybe to his friends is funny coz he's that sort of a guy.
That's not the problem. THe problem is that it got re-tweeted or forward to the police - whoever did that, did they stop to consider the original intent. Or was this thoughtless? or possibly even malicious? why did the police not regard this as potentially a "crank call"? were they not wasting police time

2/ research on hypothermia received a boost from the Nazi doctors trying stuff out on people in concentration camps. The research actually turned out to be useful. Should we deny ourselves the benefit, despite the awful provenance of the results?

Some of this can be reduced (in a classic reductionist manner) by looking at cost-benefit tradeoffs. What are the risks the tweet really is a bad guy? what are the chances that if we use the nazi medical results, someone will start a land war with russia, to aid in their drug discovery programme?
How much extra work do we have to do to make deciding the right way about the incentives, or the re-use of results?

Why do I care?

Well, its clear that many open, public-minded people constrain themselves from doing potentially valuable research by setting barriers before working out the cost-benefit/incentive balance, whilst at the same time, a lot of industry just goes ahead and does it, and fixes things up after the event.
Is there a middle ground?

I'm suggesting that considering re-use and the safeguards one should put in place for that, might give a way to evaluate whether something is ethically acceptable or no.

It could also serve by offering "re-use cases" which might be easier to explain to people as part of the "informed consent" steps of any ethical experimental design.

As more pressure is put on us to do work with larger and larger data sets (whether the GCHQ surveillance data, or police crime/geo-loc data, or NHS care.data) we need to figure this out soonest.

Sunday, January 25, 2015

Intellectual Property and Production - workshop in Law, 24.1.2015

at Wolfson College, 25 Jan 2015
Organised by Center for Intellectual Property and Information Law

Jennifer Davies introduced the workshop (6th in a series run by CIPIL ) and reminded attendees this was about production (and place!)

Session 1: Producing Output, led by Lionel Bently

Tanya Aplin: 'The Impact of Prosumers on Photojournalism' - this talk mapped out how the massive rise of amateur photography and the ubiquitous presence of cameras (especially in phones) had given rise to the de-skilling of the area of photojournalism, starting especially in disasters and emergencies (Katrina, Tsunami, Arab Spring etc), where classical photojournalists might not be on the scene for days in any case, but now supplanting every day coverage of events. The talk then went on to the new businesses that had arisen to "professionalise" the contributions by the public, and new agencies that would find takers for pictures and pay producers (albeit massively less than the famous photojournalists of yore).

Great talk - i think to be fair, the "professionals" always over-stated the skills you need to be a good journalist - many amateurs are massively better than many paparazzi - see my friend tim harris photos - he's a research leader at Oracle labs....this is also true of bloggers - for example my local kentish town blogger produces more timely and better job than the main traditional local newspaper (in my view:)

Alan Durant: ‘Kinds of Work in Copyright: a Semantic Perspective’

This was a foundational take on the triptych "Original Literary Works" unpicking the individual and combined meanings over history and showing how ambiguous the terms sepeaately and together were and still are. THis has important consequences for public and other stakeholders understanding of the relevant law!

Hye-Kyung Lee: ‘Media fans as new cultural intermediaries’

This was about fan contributed works with a new spin - the Manga translators - fandom who take japanese and korean comic art, and produce versions for other markets! Again (as with the first talk in this session) the emergent business models have first been squashed, but then embraced and extended by the original Manga production houses. Fans reactions vary (as you would expect) from approval to "uncool" - also, interestingly, cultural differences mean that Chinese reaction is quite different than other countries in this regards. I asked if anyone has tried personality profiling (c.f. psychometrics facebook app in cambridge) to see what sort of people enjoy these activities - is it mostly about fame, social capital, etc - inclusion - etc or do some really want to make it like their heroes in the original comics? or a mix (changing over time). Was reminded of two related stories. (Anecdotal) - 1/ The Dr Who fan contributed plots and characters were threatened with a lawsuit for copyright reasons, but then it emerged some of the writers of Dr Who had been reading the site, and may have inadvertently used material and fans suggested that they might countersue - so a quid pro quo was reached:) 2/ My cousin was teaching computer science in Brasil and translated US text books by reading then speaking aloud in portuguese, to a dictation/speech to text system (Dragon Dictate) - students bought the original book, so there's no loss to publishers or authors, but were given handouts of the text part in their own language.  Publishers became interested (low cost way to get foreign language editions of books....and discover markets).

This element of fan led A&R seems like an obvious way for producers of new work who live in the "long tail" of the popularity distribution to find their market, and for new fans to find content. Seems like a good thing - for a given area, it may not be a zero sum game  although over all content, I guess the population does have a finite disposable income. However, this new mode may lead to a wider breadth of availability, perhaps at the expense only of the few people at the very tip of the popularity (U2, Beyonce might lose a few percent of their gazillions) - maybe Thomas Pikety will approve!

Session 2: The Means of Production (I) led by: Ian Walden

Jon Crowcroft: ‘Software Tools and Die’
here's my talk which is mostly just about how we write software but also a plea not to do s/w patents (the clever bits are just math) and how copyright is ok, but there's lots of different uses (license models) which can all work (and co-exist)

Catherine Seville: “Controlling the Production of Literary Works: Hard
Then, Impossible Now?

This was a fascinating talk about the ways in which the technologies for production have always seen (mostly failed) attempts to control - from limited access to printing press, to control of rights to publish, through to DRM etc - really interesting...

Simon Eliot: ‘From 'Literary' to 'Commercial' in the Registry Books of
the Stationers' Company

Beautifully read and presented talk about this history and hilariously incompetent antics of rhe Stationers' company (registry for all things trademarkish) esp in the 19th century, which some fabulous graphic examples of Victorial graphic design for a wide range of useful, daft, and sentimental works. Sheer diversity of display is astonishing.

Session 3: The Means of Production (II) led by: Andrew Griffiths

Sean Bottomley: ‘Trade secrecy and production during the late
seventeenth and eighteenth centuries.’

Lovely talk about Worshipful socieiies and their control of their magic pixie dust techniques-  interestingly, the use of sanctions against people for revealing these tricks to "foreigners" (i.e. non guild members) was relatively rare in practice.  It makes you wonder why the Worshipful Company of Information Technologists was set up - What have they got to hide???

Dev Gangjee: ‘Why Wine is Not a Commodity: Geographic Indications,
Place and Process’

Great talk as with all the talks, really well presented and clear) on "terroir, and the way GIs are being used more and more, and yet have very little public recognition yet (outside of the original "Appelation Controle) - I asked the questions a) how "little britain" is all this and b) can you have a GI for "produced near here" for any definition of here (i.e. eco-didn't fly far food) - answer, yes...

Theresa Lopes: ‘Trademarks and Production of Consumer Goods from an
Historical Perspective’

Great econometric analysis of a large body of data from several places showing how IP protection trends track a) innovation b) the economic growth/shrinkage - a lot to take in - hopefully, we get some slides/paper post workshop!


Session 4: Geographical Delineation, led by Session leader: Graeme Dinwoodie

Sun Thathong: ‘A Marxian interpretation of the traditional knowledge
debate: the role of IP in the
coexistence of primitive and capitalist modes of production.’

Fun stuff -not quite as alarming as it sounds - basically, traditional knowledge is folk lore and crafts embedded in simple societies and held in common. It is being ram-raided by developed countries profit led companies (most obviously, plant medicines in rain forests being patented by big Pharma). This is about the moves to try and figure out a range of IP models that will appropriately recompense those "simple" societies (which, if course, are never so simple). Main bug is the hilariously naive Marxian description of primitive communism, which an anthropologist would have a field day with (literally and figuratively), but the model economically is good.

Here I had to leave to go to a memorial Mass for my brother in law, Declan McKeever
so I missed one of the organisers papers, which was a shame as it was
foundational to the main purpose of the workshop:

Jennifer Davis: ‘The End of Trade Marks as an Indicator of Place and
the Implications for Labour’

Graham Dutfield: ‘The compulsory working of patents and developing
countries.’

Henning Grosse Ruse-Khan: ‘From Local Manufacture Rules to Import
Barriers in Global Production Chains’