# Adaptive secure data transmission method for OSI level 1

Chapter I

1. Introduction

This chapter is organized into background, motivation of the work, evolution of telecommunications,

objectives and, outline of the thesis. The motivation includes earlier activities and work

experiences and contribution papers of this thesis. The evolution section gives a review of

other studies of telecommunications, standard modems and their performance, waveforms and

the idea of multi-carrier systems. Research methods and research problems are discussed in the

last section: objectives of the thesis and outline of the work. The objectives describe the aim of

the research. Formulation of the adaptive secure data communication method for OSI level one,

its potential applications and new knowledge of the use of discrete Fourier transform (DFT) in

generation and detection of waveforms is the expected benefits of this study. The outline of the

work defines the structure of the thesis.

1.1. Background

The most unfortunate things in the modern world awoke the world, the World Trade Center on

September eleven in 2001 and an earthquake followed by a tsunami in Asia on Boxing Day

26.12.2004, when we think of all the precautions against manmade catastrophes and convulsions

of nature. The former perfected a military transformation in USA. The latter made it clear

that an information system based on telecommunications could have saved most of the nearly

180 000 deaths. It is sad that a working alert system was not made there, in spite of the technology

known and used in most countries in the Pacific Basin. The Pacific Tsunami Warning Center

(PTWC), established in 1949, provides warnings for teletsunamis to the areas and US interests.

The communication system from the information source via radio waves (public broadcast,

mobile cellular, satellite service etc) to the people exists. The completing alternatives to public

broadcasting, for example messages with cellular phones, can serve as redundancy in securing

the authority made alerts. The motto of the Signal Regiment of the Finnish Defence Forces

(FDF) says “Denuntiatio solum translata valet”. The tsunami also reminds the author of the development

efforts within telemedicine in general as one of the most important issues. It is worth

all the efforts already made and those that will be made in the future. This encourages the author

in publishing some results of the secure adaptive data communication methods studied and

developed during several years and especially during 2000-2005 at the Tallinn University of

Technology. The results of this thesis will be presented on the following pages. An adaptive

data communication method (soft data generation and DFT detection method) will be presented

that can be included in public broadcasting systems all over the world. The idea is that broadcasting

stations are coupled in the alerting data sources and local authorities have standard receivers

equipped with a small additional audio circuit for detection of an alerting coded waveform.

For telemedicine communications the waveform will be proposed as a secure OSI level 1

wireless data transmission.

11

1.2. Motivation of the Work

The public switched telephone network (PSTN) in Finland was in an early phase of the automatic

traffic during the time 1968-1970, when the author was an office engineer at the Post and

Telegraph Administration in Helsinki. The value of automatic telephone networks was well understood

during the automation projects in various places in Finland (Porvoo, Imatra, Keuruu,

Lieksa, Kristiinankaupunki, Saarijärvi, Kemijärvi etc). The author completed his M.Sc thesis in

1969 about the R2, which is a MFC signaling in the automatic PSTN. The result of that work is

described in reference [Par00].

During active military service in 1970-1977 the author was also involved in discussions

concerning the automation of the, at that time, manual tactical military telephone services

[Lei72]. The author’s first post-graduate thesis (Lic.Tech 1975) and a publication

in HUT (1987) was a result of the studies for the evaluation of the availability of a limited

network, papers [Lal75] and [Lal87]. For e_{leven years, 1977-1988, the author worked}

at the vocational training centers at Jyväskylä leading a department and at Järvenpää as a

leader. Among several activities during that time he was also involved in the development of

training programs for numerically controlled (NC) machinery and microcomputers. At that time

the evolution of microprocessors and personal computers (PC) was intensive. The use of PCs in

education and small businesses with the idea of electronic sheet, Visicalc in 1978-1979

[Rad80], was started. The author also had the TRS-80 Model I at home.

The post of Assistant Professor of Applied Electronics at Lappeenranta University of Technology

was founded in 1989 and the author was its first holder conducting courses on “Microprocessors”

and “Machine language programming. The author published a textbook about microcomputers

used in vocational training centers and Lappeenranta University of Technology. At

Helsinki University of Technology the author has conducted courses on “Teleautomation” in

1984-1985 and 1988-1989, and “Digital systems” in 1990-1991 working at the Department of

Electrical and Communications Engineering.

The automation and digitization of military networks, both tactical (moving) and strategic

(fixed), and the start-up of the military information age with the Intranet and first personal

computers (PC) was established in about 1988. This time was his second military service in

1988-2004 (retired on 31.12.2004). The author designed training guides and planning methods

for new systems including modeling and simulation methods described in references [Cac93,

Mis98] and a robust simulation tool of his own [Lal04a-b]. Design methods for the tactical radio

links and wireless networks were one of the main efforts during 1998-2004. The network

planning system is based on modeling and computer simulations using cartographic data from

Finland (digital maps, elevation models, clutter and forest height information).

Oracular forecast is found in reference [Fdf03] about the roles and tasks of defense forces and

their mutual relations. International peace keeping forces are gaining a more significant role as

the threat of armored attack diminishes. Large-scale national disasters and defense against terrorism

necessitate the use of armed forces in support of Civil Society. The reference presents a

software radio program and its new adaptive wideband networking waveform for FDF [Fdf03].

Also the need of command and control (C2) systems with wireless communication networks

using (VHF, UHF) digital radios motivated into several studies of adaptive data communication

systems One result of the author’s studies in these areas was the concept of an adaptive modem

and later an adaptive security method for data communication [Lal99, Lal00, Lal02, Lal04b].

The adaptive modem was found as a novel solution in the PCT review in 1999. Both the adaptive

securing method for OSI level one (physical) data transmission and the adaptive modem itself

have been awarded with an invention prize of the FDF in 1999 and in 2002. This system is

the main result in this thesis, discussed in several references [Lal97a-b, Lal99, Lal00, Lal01,

Lal02, Lal04a-b].

12

The background described above gave the motivating force for the academic work at the Tallinn

University of Technology during 2000-2005 with the results published in the following proceeding

of

- Eurocom conference 2000 in München “Adaptive modem” [Lal00].

- MILCOM conference 2001 in MacLean, USA with title “Adaptive modem techno logy”

[Lal01]

- MILCOM conference 2002 in Anaheim, CA, USA with title “Basic theory of adaptive data

transmission” [Lal02].

- ESM conference 2004 in Magdeburg, Germany with title “Modeling and simulation of

biomedical data networks” [Lal04a].

- MILCOM conference 2004 in Monterey, CA, USA with title “Robust simulation of wav e-

forms” [Lal04b].

The main contributions of this thesis are in the theory of adaptive data communications, a simulation

of telemedicine traffic, and the soft detection and generation method of adaptive bandlimited

waveforms based on the discrete Fourier transform (DFT), Appendix 1. The result of the

studies is defined in the proposal for secure telemedicine and biomedical information data

transmission method for OSI level 1.

1.3. Evolution of Telecommunications

A Shannon’s result is the concept that every communication channel had a speed limit, mea s-

ured in binary digits per second. The famous Shannon Limit is the familiar formula for the capacity

of a white Gaussian noise channel [Sha48]. In a paper [Mos02] a new paradigm called

“capability" is proposed, which gauges the effectiveness of a steganographic method. It includes

payload carrying ability, detectability, and robustness components. A demonstration is made

that a compressed image (JPEG) always has the potential to carry hidden information [Mos02].

In the adaptive data transmission method studied in this thesis hidden information will be carried

in the signal space. The use of zero-error capacity for channel analysis is discussed. The

error free channel capacity is one substance used in this thesis.

1.3.1. Studies of Telecommunication

Alec Reeves invented PCM in 1937 as a result of a research group in Paris [Alc92, Rob04].

Severe problems with noise and distortion were reported while bandwidth was large in radio

links. Technologically it was too early to use PCM in practice. The encoding of signals with 1-

bit code has been studied quite early 1946 [Del46]. The introduction of delta modulation (DM)

by de Jager was made in 1952 [deJ52]. Voice coding methods (PCM and DM) produce granular

noise. Bandwidth and signal-to-noise ratio (SNR) are important parameters used in evaluation

of a system’s pe rformance in this thesis.

Voice Grade Data Communication

At present, “capacity" is the prevailing paradigm for covert channels. Voice grade data tran s-

mission on wired channels has been studied by ITU-T as early as in March 1960 [Itt61, Itu60].

Modulation method of the first data transmissions was FSK and later MFSK. Modem development

is based on modulation method, bit rate and symbol rate, Table 1.1.

13

Table 1.1. Standard modems [Lal97a]

Modem development is seen in the modulation method: phase difference, phase and amplitude

difference (QAM), and finally coding methods were used. The key idea was that the operations of

modulation and coding are combined. Limitations on conventional block and convolutional codes

on band-limited channels have motivated to introduce other coding methods TCM (Trellis-Coded

Modulation) scheme, discovered by G. Ungerboeck in 1976 [Ung82] and its several variants for

example iterative turbo BICM (Bit Interleaved Code Modulation) scheme [Big91 SNg03] for decreasing

error probabilities.

Performance Analysis

The performance analysis for a TCM system with a Viterbi decoder [Vit67] is derived in reference

[Ung82] as

The upper bound on the node error rate is

P

e

⎛ ^{2}

d E ^{⎞}

free s

⎛ ^{2}

d E ^{⎞}

⎜ ⎟

free s ^{⎜ }^{4}^{N }^{⎟}

⎝ ^{0 }⎠

^{≤ }Q^{⎜ ⎟}e T (D) ^{(1.1)}

⎜ _{2}_{N }⎟

0

⎝ ⎠

with

⎛ _{E }⎞

⎜ ^{s }⎟

⎝ ^{4}^{N}^{0 }⎠

=

D e _{(1.2)}

14

and the upper bound on BER is:

P

e

⎛ ^{2}

d E ^{⎞}

free s_{2 }⎜ ⎟

1 ^{d}

free^{E}s ^{⎜ }^{4}^{N }^{⎟}

⎝ ^{0 }⎠

⎛

≤ Q

⎜

m ^{⎜}

⎝

2N

0

⎞

⎟_{e}

⎟

⎠

∂(D, I )

∂I ^{(1.3)}

with

⎛

⎜

⎝

E_{s}

4N

⎞

⎟

⎠

0

D = e _{, }_{I=1 }_{(1.4)}

Where ^{T(D) }is the generating function of the direct graph (state diagram of the trellis code).^{T(D,I) }is the augmented version of ^{T(D)}with components of the I terms denote the number of information

bits errors associated with each error event. ^{N}_{0 }is the one sided noise spectral density

of AWGN, ^{E}_{s }is the average energy of the signal constellation and m is the number of information

bits carried by each symbol.

The derivation of generating function is complicated as the number of states in the TCM increases.

At high S/N values an approximation is quite accurate without ^{T(D,I) }[Ung82]:

⎛ ^{2 }⎞_{⎜ }^{d }free^{E}s

P ^{≈ ⎟}

e ^{N}^{(}^{d }free ^{)}^{Q }^{(1.5)}

⎜ _{2}_{N }⎟

0

⎝ ⎠

d

P ≈ _{m}

⎛

Q^{⎜}

⎜

⎝

d

E

⎞

⎟

⎟

⎠

2

b

free free s

2N

0

(1.6)

Here the ^{N(d}_{free}^{) }is the average number of sequences that are distance ^{d}_{free }from the transmitted

sequence.

For example, using TCM the error rate can be reduced by three orders of magnitude (from 1 in 10

to 1 in 10000). In a classical view the spectral efficiency as the number of bits per second transmitted

per one hertz of bandwidth is decreased due to the increased code rate. However, the

modulation symbol set could be enlarged when coding is used relative to that needed for uncoded

case [Ung82] and:

- If the signal set dimensionality per info bit is unchanged the power spectrum remains unchanged

(no BW expansion or change in spectral efficiency).

- The signaling rate does not change if coding and modulation is performed with respect to ED

(Euclidean distance) between coded modulation sequences.

TCM code is recommended by ITU-T with

V.32 for 9.6 kbps over two-wire telephone lines and 14.4 kbps one-wire lines.

V.17 for use with 64-QAM and 128-CROSS 14.4 kbps FAX traffic over standard phone lines.

V.33 14.4 kbps and V.34 28.8 kbps.

For evaluation of a modulation method a formula for BER versus S/N (bit error rate versus signal-to-noise

ratio) are generated.

15

In general the Gaussian error integral, function

^{Q(x}), gives an approximation for error performance

^{P}

_{e }of sequences of transmitted symbols in signal space, where

^{d}

_{min }minimum Euclidean distance

described in references [Vit67, Ung82]. At high signal-to-noise ratios [Ung87]:

dy

e

x

Q

x

y

∫

∞ −

=

^{2}

2

2

1

)

(

_{π }

_{(1.7)}

⎟

⎠

⎞

⎜

⎝

⎛

=

_{σ}

2

min

d

Q

P

e

^{(1.8)}

⎟

⎟

⎠

⎞

⎜

⎜

⎝

⎛

=

_{σ}

2

free

free

e

d

Q

N

P

^{(1.9)}

Where

^{N}

_{free }is the average number of sequences at

^{d}

_{free}. In TCM the Euclidean distance

^{d}

_{free }between

signal sequences is increased to get a lower error rate. The Euclidean distance between two

points

^{p }and

^{q }in N dimensions

^{i=1…N }as

∑

=

−

=

N

i

i

q

p

d

1

2

1

^{)}

(

^{(1.10)}

Formulae (1.11)-(1.12) for FSK and DPSK are proper references for other modems developed

later, reference [Car86].

2

2

1

^{γ}

−

=

_{e}

P

b

^{(1.11)}

For FSK, envelope detection, and modulation speed

_{1}

/

^{=}

T

b

^{B}

r

^{.}

γ

−

=

_{e}

P

b

_{2}

^{1 }, (1.12)

For DPSK, phase-comparison detection and modulation speed

_{1}

/

^{=}

T

b

^{B}

r

^{.}

A comparison of modulation methods is presented in reference [Car86] pp. 553-554. The selection

of a modulation method is made at a common standard for comparison purposes

4

10

^{−}

=

b

P

_{.}

For example, M-ary DPSK is a reference for modulation methods, formula (1.13). The performance

of 2-PSK…32-PSK is at

4

10

^{−}

=

b

P

_{about }5

...

1

/

^{=}

T

b

^{B}

^{r }bps/Hz and

^{21}

...

8

=

γ

_{dB.}

⎟

⎟

⎠

⎞

⎜

⎜

⎝

⎛

=

_{M}

K

Q

K

P

b

_{2}

sin

4

1

_{2 }

_{π}

γ (1.13)

Where bit error rate is ^{P}_{b }and χ is energy to signal ratio, data conversion factor is

K ^{= }_{log}_{2 }M ^{K}

and number of symbols in a M-ary system, for example ^{M }^{= }^{2 }.

In this thesis FSK and DPSK are used as references in performance analysis.

Waveforms

The basic waveform is a simple sinusoidal signal as presented in formula (1.14)

s_{(}t_{) }^{= }A_{sin(2}^{π}

mft ^{+ }P_{) }_{(1.14)}

It has four information carrying parameters in it (A, f, t and P). These parameters are used in

data modulation methods. The multi-carrier or multi-channel signals are described Waveforms

as a Fourier presentation in formula (1.15)

∞ ∞

1

S(t) ^{= }c ^{+ }_{∑ }a_{n }sin(2πnft) ^{+ }_{∑}b_{n }cos(2πnft) ^{(1.15)}

2

n_{=}1 n_{=}1

Any signal can be presented as a Fourier series [Mar62]. An algorithm for the machine calculation

of complex Fourier series was first presented in 1965 [Coo65].

Digital Generation of Waveforms - A symbol-based approach

In present digital networks data transmission is virtually transmitted in digital form. The nature

itself is analog and also analog are the waveforms of Internet access circuits and radio waves

(presently called air interface, former name was ether). Thus the need for analog waveforms

and modeling methods still exists. However, the waveforms themselves can be made with digital

modulation methods. The present studies of waveforms are focused on wideband applications

(B-ISDN, UMTS, WNW etc) both in fixed wired and moving wireless systems. Several

applications are based on the OFDM waveform (Orthogonal Frequency Division Multiplexing)

scheme. The basic idea of multi-carrier modulation was introduced and patented in the mid 60s

by R.W.Chang [Cha66, Wei71]. The sampled waveform may be presented as x(t) in [Guo02]

formula (1.17). The analog signal at DAC output after quadrature modulation in a software radio

type modulator is

~ ^{))}

j_{(}_{ω}

C^{t}^{+}^{φ}

(^{t}

s (t) ^{= }R{x(t)e } ^{(1.16)}

With amplitude of the input signal to the RF transmitter

+∞ ^{N}_{S }^{−}^{1}

2 2 ^{T}_{S}

x(t) _{∑ ∑ }{[s_{I}

(k) ^{+ }s_{Q }(k)]p(t ^{− }k ^{− }iT_{S })} ^{(1.17)}

N

=

i_{=}0

k_{=}0

17

and the phase

~_{s }_{(}_{t}_{)}

−^{1 }^{Q}

φ

(t) ^{= }tan ( _{~ }) ^{(1.18)}

s (t)

I

where TS is the period of one time-domain OFDM symbol, NS is the number of samples in one

symbol after CP and p(t) is the pulse function of the symbols. The high peak-to-average-power

ratio (PAPR) of OFDM systems introduces non-linear distortion in the transmitter and causes

both in-band and out-of-band spectrum re-growth. Undesired effects are [Guo02]:

- Nonlinearities.

- Intermodulation among subcarriers.

- Undesired out-of-band radiation.

In general, the complexity of the equalizer grows with higher symbol rates because as the transmission

rate increases (symbol periods become shorter), the effects of ISI degradation are higher.

In the case of both DMT and OFDM, the channel is divided into many sub-channels situated in

different frequency bands, with each sub-channel utilizing a carrier with Quadrature Amplitude

Modulation (QAM). This allows a bit stream with a very high transmission rate to be sub-divided

into many bit streams with lower transmission rates, reducing the relative effect of ISI in each

symbol period (which is now longer) and allowing simplification of the equalizer. Therefore, both

OFDM and DMT systems can implement equalization in a simple fashion, due to their slower

symbol rates. The OFDM waveform is resistive against multi-path errors.

In this thesis band-limited multi-carrier waveforms are the main subject.

Channel Modeling

References of channel modeling [Sha48, Rum86, Agu03] and recommendations [Eur86, Itu88]

give a classification of channels:

AWGN used for theoretical modeling and reference in evaluations of modulation methods.

Granular noise channel is generated in the digitization process of analog signals.

Multi-path channel is a wireless (mobile cellular, radio broadcast) channel.

This classification is used in this thesis.

1.3.2. Evolution to Adaptive Communications

Adaptation is known as an adjustment by which a species improves its condition in relation to its

environment. Culture is the human adaptive system.

Idea of Multi-Carrier Modulation

The concept of Multi-Carrier Modulation (MCM) or Discrete Multi-Tone (DMT) has been known

for many decades. In the 1960s, a multi-carrier modulation technique known as Orthogonal Frequency

Division Multiplexing (OFDM) was invented, which utilized multiple sub-channels in the

frequency domain [Cha66, Wei71]. The basic idea of multi-carrier modulation was introduced

and patented in the mid 60’s by R.W.Chang. The channel is sliced up into narrow little bands and

multi-carrier modulation is used with individual modulation in each of the bands [Gal68, Gal01].

MCM techniques (OFDM and DMT) have evolved and have been adopted in various standard

bodies such as IEEE 802, American National Standards Institute (ANSI), European Telecommunications

Standards Institute (ETSI), and International Telecommunications Union (ITU). Multi-

Carrier Modulation (MCM) [Sta99], divides the data into a number of low rate channels that are

stacked in frequency and separated by 1/symbol rate. MCM, also called OFDM, is being proposed

for numerous systems including audio broadcasting (DAB), video broadcasting (DVB),

mobile wireless access (WLAN) and digital subscriber link systems (xDSL).

18

Definition of OFDM

The recent OFDM method (sometimes called multi-carrier or discrete multi-tone modulation) is

the basis of several standardized systems for data networks and cellular radio communications. A

definition of OFDM systems [Eng03] describes it as: The available bandwidth W is divided into a

number Nc of subbands (subcarriers or subchannels) each of width f=W/Nc. Data symbols are

transmitted in parallel by modulating the Nc carriers. To assure a high spectral efficiency, the

subchannel waveforms must have overlapping transmit spectra. They need to be orthogonal for

enabling simple separation of these overlapping subchannels at the receiver. Multi-carrier modulations

that fulfill these conditions are called Orthogonal Frequency Division Multiplex (OFDM)

system.

OFDM System Model

A high-level block diagram [Ram02] is shown in Appendix 2. The system model of an OFDM

transmitter with RF is presented in reference [Guo02]. The information bits

[b^{1 }b^{2 }b^{3 }… b^{M}] (1.19)

are mapped into the I/Q channel baseband symbols using a modulation scheme such as phaseshift-keying

(PSK) or quadrature-amplitude-modulation (QAM). Then each group of N symbols

are packed into a parallel block

[ ]^{T}

S S S _{⋅⋅⋅ }S

1 2 3_{⋅ }N

(1.20)

at the input to the IFFT.

OFDM symbols in the time domain over time interval

[_{0 }_{T }]

t ^{∈ }, ^{(1.21)}

S

are generated by the IFFT operation as,

s(k) ^{= }^{1}

N

N

∑

n_{=}1

S

j k_{− }n_{− }N_{n}_{e }2π( 1)( 1) /

(1.22)

for ^{k=}^{[}^{1,2,…,N}^{]}

Then in Cyclic Prefix (CP) insertion, the first G coefficients are repeated after the original N coefficients

and made serial for quadrature modulation. The analog signal at DAC output after

quadrature modulation in a software radio type modulator is,

~ ^{))}

j_{(}_{ω}

C^{t}^{+}^{φ}

(^{t}

s (t) ^{= }R{x(t)e } ^{(1.16)}

with amplitude of the input signal to the RF transmitter

N 1

+∞ −

x(t) ^{= }_{∑∑}

i_{=}0 k_{=}0

{[s

2

I

(k) ^{+ }s

2

Q

T_{S}

(k)]p(t ^{− }k _{N}

− _{iT}

S

)} ^{(1.17)}

19

and the phase

~_{s }_{(}_{t}_{)}

−^{1 }^{Q}

φ

(t) ^{= }tan ( _{~ }) ^{(1.18)}

s (t)

I

Where TS is the period of one time-domain OFDM symbol, N is the number of samples in one

symbol after CP and p(t) is the pulse function of the symbols.

1.3.3. Discussion of Broadband Evolution

The evolution to adaptive communications is due to many different factors including the development

of MCM in the 1960s. Advances in the design of new communication systems are

based, later in the 1970s, on the microprocessor, software memory, digital signal processing and

chip manufacturing technology. Methods for doing coding, waveform shaping and equalization

are still a developing area in communications. The terms “multilevel QAM digital radio sy s-

tem” or “variable rate QAM” or “symbol rate controlled modulation” were used i n the late

1980s. Self-adjusting adaptive systems and OFDM (Orthogonal Frequency Division Multiplex)

systems have been designed and used for many civil applications in the 1990s. The recent ubiquitous

OFDM method is the basis of several standardized systems for data networks and cellular

radio communications. In the 1990s also a new concept “adaptive modulation” is used.

Modeling and simulations have been used for telecommunication network planning since the

development of simulation programming and the first simulators in 1964. Simulations are important

in the evaluation and design of network structure, communication routes with traffic

calculations, link budgets, radio wave propagation and circuit level analysis. At about the same

time an algorithm for the machine calculation of complex Fourier series was presented in 1965.

In the 1970s Discrete Fourier Transform (DFT) was studied. Soft generation and detection of

waveforms with IFFT and FFT, Object-Oriented Analysis (OOA) were used in the 1990s.

Military applications will use new downloadable waveforms (a software algorithm) with Software

Defined Radios (SDR). WNW (Wideband Networking Waveform) waveforms will be

designed for military mobility and network access purposes according to some military development

plans during 2004-2007.

1.3.4. Discussion of Some Communication Problems

Biomedical Data Transmission Problems on Physical Level

Biomedical data transmission problems on the physical level of the OSI model are presented

according to reference [Var03]:

1. For the purposes of various databases and information systems off-the-shelf solutions are

available. There are already many standards for platform independent representation and interchange

of static or quasi-static medical records (patient data, insurance data, medical images

etc.). These communication standards are based on the upper layers of the OSI model,

and the lower layers, like the physical transmission, the media access and the routing methods

are not defined.

2. As it can be established, to fulfill all of these requirements we must define the lower layers

of the OSI model. Although many vendors are offering their monitoring systems, they lack

an open communication technology and so lack the ability of interchangeability between

devices of various vendors.

20