Fundamental Limits vs. Limits to Imagination

1
Fundamental Limitsvs.Limits to Imagination

• October 2000

2
The General Outline

• Frame the general issue
• Provide some background
• Discuss some specific technical issues

3
Framing the Issue

• Our apparent motivation
• To encourage the search for fundamental technical
  limits in our ability to do signal processing
• Our deeper motivation
• Those limits, and the rate at which were
  approaching them, encourage/discourage RD,
  investment, and, ultimately, educational
  approaches
• A closer examination shows that psychological and
  financial limits are just as important as
  technical ones

4
Background

• Well known examples of technical limits and
  trends
• Shannon
• Moore
• The impact of human psychology
• Faith
• Tolerance of risk
• The appearance of the third way
• The emergence of financial limits
• The move from a theory to an application-enabler
• The end of the era of the electromathematician

5
Well Known Examples

• Moores Law (an observation, actually)

000306

• General description Factor of two improvement
  in semiconductors every 18 months (since 1965)
• The big question (1) will it continue?, (2)
  where does it end?

6
Moores Law
7
Well Known Example 2

• Shannons Capacity LimitC B log2 1S/N

• Application to 4-wire telephone transmission
  channels
• Existence of easily quantified limit encouraged
  both theoretical research and product development

8
Impact of Human Psychology

• Claim Observed trends and the existence of
  fundamental limits gives a way to judge the
  viability of a research, development, or
  educational endeavor

• Second claim Not when humans are involved!
• Age dependence in acceptance of trends and limits

9
Putting Your MoneyWhere Your Analysis Is

• Fiber-based transport is getting cheaper much
  faster than silicon-based computation or radio
  transmission
• Implication The cheapest network will use fiber
  transport to avoid switching, computing, and RF
  links

10
Cheaper Fiber Transport is Altering Telecom
Network Topologies

• Important implications
• Signals will be concentrated into fewer physical
  paths
• Electronic switches will be fewer and larger
  networks will be flatter
• RF end links will be as short as possible

11
Are You Ready?

• Will you believe it, and are you ready to act on
  it?

12
The Third Way

• The Third Way usually achieved by changing the
  rules

• Second example V.90 modems
• Wildcards often induce the third way

13
Wildcards

• Change of consumer taste
• Examples1. Is reliability really as important
  as telephony people    believe?
• 2. Is voice quality important?
• Nationalization of commercial telecom assets
• As was done to the multinational oil companies

14
Wildcards (continued)

• The Big Lie
• Examples1. PCS isnt cellular2. Digital always
  sounds better3. Cable modems have contention
  problems     but DSL doesnt
• Third-generation cellular
• Will people pay for data? They dont now!
• Will IP-based 3G forever leave behind the
  circuit-switched legacy?

15
The Emergence of Financial Limits

• Norm Augustine Using the B2 bomber as a
  financial guide, the whole U.S. military in 2020
  will consist of only one (very expensive)
  airplane
• Pentium N Not many companies can afford the
  expense of a new, world-class, state-of-the-art
  product

16
The End of the Era of Electromathematicians

• A chronology
• 18501940 Practical electricians
• 19301950 Terman and the Bell Laboratories
• 19501985 The age of linear models and white
  noise
• 1985gtpresent Computers, tools, and simulators
• 2000INFINITY Moving DSP education to the 11th
  grade
• Issues
• Observed erosion in organizational and analytical
  skills
• Where should our curricula be going?
• What skill sets will define the next generation
  of products?

17
Part III Limits in Specific Technologies

• General format
• Is there a theoretical limit?
• On what does it depend?
• Can a trend toward it be predicted?
• Does it matter?
• Is there a way out of it?
• Would you bet on it?

18
1

• How far can silicon semiconductors go?
• Whats the most computation that we might have at
  our disposal?
• What is the basis of this limit?
• What would make the limit go away?

19
2

• Is computation really the limit to our ability to
  do DSP, or is it the ability to convert the
  analog world into accurate digital
  representations?
• What limits that?
• Are there alternatives?

20
3

• What is the Shannon capacity of a pair of twisted
  wires?
• Or does it matter?Is crosstalk the real
  practical limit?
• Should we expect terabit Ethernet?

21
4

• Why do blind equalizers take more data to
  converge than do directed equalizers?
• What is the key factor?
• Is the slow-down factor predictable?
• Would the answer change if computation were free?

22
5

• Is there a theoretical limit to the number of
  erlangs/meter2 that a wireless system can
  achieve?
• If not, what is the practical limit and when do
  we hit it?
• Is the anywhere-anytime mobile wireless concept
  the wrong approach?

23
6

• Much work is going on regarding the use of many
  antennas for both the transmitter and receiver in
  a radio system
• Is the capacity promised actually achievable?
• Is the additional cost worth the additional
  performance?

24
7

• How good can batteries be?
• When will we get there?
• Is battery technology the real limit to anything
  we plan on doing?

000311
25
8

• How much bandwidth can a single human being use?
• At that rate, what would the network look like
  to provide it to 300 million Americans?
• Would they be willing to pay for it?
• Will tethers be tolerated, even if only for power?

26
9

• Does peer-to-peer (e.g., Napster and Gnutella)
  spell the end of IP protection?
• When will customers become comfortable paying for
  the marginal bit?
• Will network providers sell bits when they can
  get ten times as much for voice?

27
What Does All This Mean?

• What should we be studying and researching?
• What should we be teaching?
• What hard problems are worth solving?
• Where should financial investments be
  intelligently made?
• Where will people be willing to spend their
  corporate and/or personal money?