CHAPTER II
ENERGY MINIMIZATION
METHODOLOGYCHAPTER 2
ENERGY MINIMIZATION METHODOLOGY
2.1 Introduction
Conservation of battery energy is one of the major challenges in portable
information devices. The management of energy becomes even more critical with devices
integrating complex. video signal processing techniques with communications To
minimize the total system energy of transmission one may design a reconfigurable
Wireless image transmission system that achieves low power by exploiting both the
variability in the image data and the multi path wireless channel. Efficient transmission of
multimedia signals from energy limited portable devices requires a iresh look at the
tradeoff’ between source compression, channel coding and transmission strategies.
Increasing the complexity of channel coding algorithm to increase the
coding gain and lowering the transmit power increases the base band processing power.
For low transmission power, a less efficient source coder, which consumes less power
and achieves less compression combine with the channel coder that add a very little
redundancy, is energy optimal [19].
Reconfigurable Digital Signal Processing (RDSP) is a good approach as a
low power technique, which exploits the variabilities in the environment. Energy saving
are achieved by tailoring the architecture to the environment [20].
46One has to find out a strategy for optimum energy consumption in
transmission and signal processing, which includes source and channel coding, and
mathematical tools to calculate efficient operating points. Allocation of energy is
required between source compression, channel coding and transmission tasks to minimize
the total energy dissipation while keeping the end-to-end distortion of the source
constant, For a given total energy budget, one has to estimate the minimum distortion
that the source encounters. Bandwidth, or transmission rate is another important system
Parameter, which aes the system performance.
Rs Pt
yt 1 ANTENNA
IMAGES —_— 7
Source Channel : f 5 ]
a encoder Encoder |,» RF & analog | ,] POWER — Le
ruin) (RS) components J] aMPLirIER [J~*] 7
— = |t
Source Channel | RF & analog x
IMAG | E 8
our} decoder fe decoder fg} RAKE components |_|
" (SPIHT) (RS) —— R
ft em
tee
@:
From source Z ;
codec OPTIMIZATION] conteuenion
atcorrimim — |_so*figuration
From channel oRLooKuP | mln)
codec
”
Fig:2.1: A reconfigurable multimedia system: (a) the transceiver and (b) the controller.
47The total energy consumed by the mobile at the link layer and at the physical layer
consists of the energy dissipated by the source compressor Es and the energy used to
transmit the compressed bits though the channel Et, resulting in total energy Etot = Es
EL joules. The methodology is suggested to minimize the total energy considering the
effect of distortion at source end and at the reception, The Fig 2.1 reconfigurable
multimedia system is shown in which transceiver and its controller is shown which is
responsible to conFig the various parameter to decide the operating point for with respect
‘o data rate and perceptual quality of the image in terms of allowable distortion.
Minimize Etot subject to Dtot < D0; (21)
Where Dtot refers to the total distortion due to source compression and channel errors,
Total allowed distortion Do is determined by the particular source and application at
hhand and is different for telephone calls, file ranster and video conferencing,
Algorithms that compress more, resulting in a better-compressed rate for
‘identical "quality" have higher complexity and require more processing energy. For fixed
source distortion Ds in Fig 2.2(a), the processing energy Es is a decreasing function of
the compressed bit rate R shown in Fig 2.2(b). It is seen that as the number of bits
representing a source increases, the transmitted energy per bit decreases. To keep the bit
crror rate of the channel constant, higher transmission energy (Et) will be required for
larger source rates R as in Fig 2.2(c), Combining in Fig 2.2(d), we find that the total
energy Etot as a function of the bit rate has a minimum and the optimal operating point
R® that minimizes Etot (in joules/source sample) for fixed end-to-end source distortion
48‘can be calculated.[21] ‘The total energy per source sample due to signal processing and
transmission is
Etot = es *N+{ RNO In (2pe) (2.2)
Where Et is the transmission energy given by
Et = -RNO In(2pe)) ————-—---—2.3)
& Esiis the signal processing energy, energy given by
Ese 05 1N nneeneenneenneenenn (4)
Where N is the transform coding dimension and Cs is the proportionality constant.
Ee-B:(souee)
a)
Fig.2.2 : Minimization of total energy for fixed end-to-end distortion,
For Et we have
49pe=1-(1-ps)'™ ..
Ps = (Dy -e07(L-p )% 2
02.0? (1-p)*- 2"
Riis the source-coding rate, NO is the noise power spectral density, variance 0”, ¢ depends
‘on the quantizer used for the transform coefficients.
As the transform coding dimension N increases, the total energy and
signal processing energy, which is the difference between total energy and transmit
energy, increases. In Fig 2.3, we consider es/NO = 10, modeling a scenario in which the
noise power spectral density NO, or equivalently channel attenuation, is low. Hence for
‘200d channel conditions, one need not to spend a lot of energy compressing the source;
low attenuation enables more source bits to be transmitted through the channel.
Dtot=0.10, os/NO=10.0
25
ost,
5 10 15 2m 2B » 8
Block size N
Fig.2.3: Graph of Block Size N Vs log (energy)
50This work is aimed at developing techniques for minimizing the energy needed to
transmit image with an acceptable level of quality and tolerable delay. Two factors that
directly relates to this objective are:
(1) The use of error concealment and
ience techniques at the source coding level
(2) The allocation of communication resources at the physical layer (such as the
transmission power and transmission raic).
Work in this area of transmitting multimedia over unreliable networks,
such as IP or cellular networks has focus on various error resilience and error
concealment techniques for minimizing the effects of losses. It can be exhibited that
maximum compression before transmission does not always provide minimal energy
consumption, especially in the case of dense sensor networks with complex signal
processing algorithms. There requires a heuristic algorithm to sclect the optimal image
‘compression parameters to minimize total energy dissipation given the network
conditions and image quality constraints & to. keep them adaptable. The parameter may
be like, quantization, transform level etc.
2.2 Optimization of Image Transmission System
Based on the present general approach to optimization of image
transmission systems with rate and energy constraints in (22), characterization is done on
the image variability in terms of the operational rate- distortion (R-D) curve (see Fig
24(a)). The R-D curve determines the (zero-error) bit-rate necessary to achieve a
particular mean-squared error distortion. Characterizing the image variability in terms of
SIthe source R-D curve enables us to use a general JSCC formulation of the tradeoff
between data and redundancy.
The power limitations of the terminal depend on whether it is a mobile or
a base station. A mobile utilizes battery power, which has limited life, whereas a base has
‘access to a stationary perennial power source with unlimited life.
It is assumed that both mobiles and bases have a fixed maximum transmit power.
Fig 2.4: Graph (a) Data rate Vs Avg distortion/pk (b)Ey/N, Vs BER
‘The configuration vectors for the mobile (m) and the base-station (b),
respectively, are defined as follows: m = [Rs; tenc; Pt; crake; tdec] ,b = [Rs; tenc; tdec];
where Rs is the source rate (in bits per pixel), tenc is the maximum number of correctable
symbol symbols per block at the channel encoder, Pt is the transmit power (or output
power delivered by the power amplifier), and crake is the configuration vector for the
RAKE receiver (crake ; i= 0 implies that ith finger is powered down). The transmit
parameters of the mobile are Rs, tenc, and Pt. The receive parameters of the mobile are
the number of protection symbols per block at the channel decoder tdec, and crake. The
92parameter crake depends on the transmit power of the other device and the channel
condition, and idec is same as ‘enc of the other device (either the base-stetion or the other
mobile), Further, M and B are the configuration-spaces (defined as set of all possible
configuration vectors) for the mobile and base-station, respectively.
Fig 2.1(b) shows the block diagram of a controller. The controller adapts
the system to the changes in the input by reconfiguring or changing the parameters of the
various blocks. The first block in the controller quantizes the input variabilities as
determined by the source coder and channel decoder. The energy-optimum configuration
vectors are either obtained by an optimization algorithm running in real-time or from a
precalculated look-up table.
In [22}- [23] dynamic algorithm transforms (DAT) were proposed as a
systematic framework for designing low-power reconfigurable signal processing systems.
DAT requires the definition of: (1) input state-space, (2) configuration-space, (3) energy
models and (4) DSP algorithm performance models. The input state-space models the
input variabilities, and the configuration-space is the set of possible hardware
configurations. Energy models and performance models are employed to obtain estimates
of the energy consumption and a performance metric (such as distortion or bit-error rate).
in order to exploit the relationship between the configuration space, the input space and
the DSP algorithm performance metric, one has to employ recent advanees in joint
source-channel coding (JSCC) [22] and DAT [23]. Significant past work on joint source-
channel coding has shown that the tradeoff between data and redundancy can be
exploited to design optimal realizable systems [24], [25], [26]. The application of.
cnt source-channel coding in heterogeneous, multi-media environments leads to general
53snatching techniques, which have been the focus of some current research [22], {27}. Jn
(221
In this chapter various JSC matching technique are discussed .These techniques
‘maximize the end-to-end quality of a transmitted image subject to constraints on the
transmit power and bandwidth A generalized methodology for energy minimization is
discussed in detailed .A matching technique which can be applied to a wide variety of
source coders and channel coders is also studied, A special attention is given to RDSP
techniques for jointly optimizing the baseband processing power and the transmit
‘power of a reconfigurable system.
34