LNCS 3141 Artificial Cellular Division by Self

Inspection 1st edition by Enrico Petraglio,

Daniel Mange, André Stauffer, Gianluca Tempesti
ISBN 3540233393 978-3540233398 pdf download
https://ebookball.com/product/lncs-3141-artificial-cellular-
division-by-self-inspection-1st-edition-by-enrico-petraglio-
daniel-mange-andra-c-stauffer-gianluca-tempesti-
isbn-3540233393-978-3540233398-10688/

Explore and download more ebooks or textbooks

at ebookball.com
 Get Your Digital Files Instantly: PDF, ePub, MOBI and More
Quick Digital Downloads: PDF, ePub, MOBI and Other Formats

LNCS 3141 Biologically Inspired Computer Virus Detection System 1st

edition by Hyungjoon Lee, Wonil Kim, Manpyo HongISBN 3540233393
978-3540233398

https://ebookball.com/product/lncs-3141-biologically-inspired-
computer-virus-detection-system-1st-edition-by-hyungjoon-lee-
wonil-kim-manpyo-hongisbn-3540233393-978-3540233398-9184/

Chemical Biophysics Quantitative Analysis of Cellular Systems 1st

Edition by Daniel Beard, Hong Qian 0521870704 9780521870702

https://ebookball.com/product/chemical-biophysics-quantitative-
analysis-of-cellular-systems-1st-edition-by-daniel-beard-hong-
qian-0521870704-9780521870702-11460/

LNAI 4335 Engineering Self Organising Systems 4th edition by Sven

Brueckner, Salima Hassas, Márk Jelasity, Daniel Yamins ISBN
3540698671 978-3540698678

https://ebookball.com/product/lnai-4335-engineering-self-
organising-systems-4th-edition-by-sven-brueckner-salima-hassas-
ma-rk-jelasity-daniel-yamins-
isbn-3540698671-978-3540698678-19678/

LNCS 2834 Pattern Classification with Parallel Processing of the

Cellular Neural Networks Based Dynamic Programming 1st edition by
Hyongsuk Kim, Taewan Oh, Sangik Na, Changbae Yoon ISBN 3540200541
978-3540200543
https://ebookball.com/product/lncs-2834-pattern-classification-
with-parallel-processing-of-the-cellular-neural-networks-based-
dynamic-programming-1st-edition-by-hyongsuk-kim-taewan-oh-sangik-
na-changbae-yoon-isbn-3540200541-978-35/
LNCS 2787 Use of an Artificial Immune System for Job Shop Scheduling
1st Edition by Carlos Coello Coello, Daniel Cortés Rivera, Nareli
Cruz Cortés ISBN 3540451927 9783540451921

https://ebookball.com/product/lncs-2787-use-of-an-artificial-
immune-system-for-job-shop-scheduling-1st-edition-by-carlos-
coello-coello-daniel-corta-c-s-rivera-nareli-cruz-corta-c-s-
isbn-3540451927-9783540451921-12196/

Epicardial Adipose Tissue From Cell to Clinic 1st Edition by Gianluca

Iacobellis ISBN 9783030405700

https://ebookball.com/product/epicardial-adipose-tissue-from-
cell-to-clinic-1st-edition-by-gianluca-iacobellis-
isbn-9783030405700-4826/

Functional Programming in C 1st Edition by Enrico Buonanno ISBN

9781638354048 1638354049

https://ebookball.com/product/functional-programming-in-c-1st-
edition-by-enrico-buonanno-isbn-9781638354048-1638354049-15788/

Markov Logic An interface layer for artificial intelligence 1st

Edition by Pedro Domingos, Daniel Lowd ISBN 9783031015496 3031015495

https://ebookball.com/product/markov-logic-an-interface-layer-
for-artificial-intelligence-1st-edition-by-pedro-domingos-daniel-
lowd-isbn-9783031015496-3031015495-19526/

LNCS 3024 Multiphase Dynamic Labeling for Variational Recognition-

Driven Image Segmentation 1st edition by Daniel Cremers, Nir Sochen,
Christoph Schnorr ISBN 3540219811 978-3540219811

https://ebookball.com/product/lncs-3024-multiphase-dynamic-
labeling-for-variational-recognition-driven-image-
segmentation-1st-edition-by-daniel-cremers-nir-sochen-christoph-
schnorr-isbn-3540219811-978-3540219811-11458/
 Artificial Cellular Division by Self-Inspection

Enrico Petraglio, Daniel Mange, André Stauffer, and Gianluca Tempesti

Swiss Federal Institute of Technology

Logic Systems Laboratory
CH-1015 Lausanne, Switzerland
[email protected]

Abstract. This article describes a novel approach to the implementa-

tion on an electronic substrate of a process analogous to the cellular
division of biological organisms. Cellular division is one of the two pro-
cesses that allow the multicellular organization of complex living beings,
and is therefore a key mechanism for the implementation of bio-inspired
features such as development (growth) and self-repair (cicatrization). In
particular, we shall describe the architecture and operation of a new kind
of programmable logic device capable of realizing, in silicon, a cellular
division process.

1 Introduction

The majority of living beings, with the exception of unicellular organisms like
viruses and bacteria, share a common multicellular organization: the organism
is divided into a finite number of cells, each realizing a single function (skin,
neuron, muscle, etc.). This architecture relies on two mechanisms:

– Cellular division is the process through which each cell achieves its duplica-
tion. During this phase, a cell copies its genetic material (the genome) and
splits into two identical daughter cells.
– Cellular differentiation defines which function a cell has to realize. This spe-
cialization, which essentially depends of the cell position in the organism, is
obtained through the expression of a part of the genome.

In a multicellular organism, each cell contains the whole of the organism’s

genetic material (the genome) and is therefore “universal”, i.e. potentially capa-
ble of replacing any other cell. In presence of a physical degradation, each living
organism is then potentially capable of self-repair (cicatrization).
Of these two mechanisms, the most difficult to realize in silicon-based sys-
tems is cellular division. In biological organisms, in fact, this feature implies, if
not the creation, at least the formation of new physical entities (cells). Such ma-
nipulation of the material substrate is not possible in today’s electronic circuits,
which cannot be physically altered after fabrication. Luckily for bio-inspired
research, programmable logic devices (FPGAs) [9,11,3] can be used to approx-
imate this process by allowing the manipulation not of the physical substrate,

A.J. Ijspeert et al. (Eds.): BioADIT 2004, LNCS 3141, pp. 217–232, 2004.


c Springer-Verlag Berlin Heidelberg 2004
218 E. Petraglio et al.

but of the logical structure of a circuit. This article describes our approach to
the implementation of a process analogous to biological cellular division in such
a programmable circuit.
Our approach is based on the Embryonics project (Section 2) for all that
concerns multi-cellular organization and cellular differentiation, and on the Cell
Matrix system (Section 3) for self-inspection based replication. Section 4 will
then introduce the system we have developed. An example of our system will
presented in section 5.

2 The Embryonics Project

The main goal of the Embryonics (embryonic electronics) project [8,10] is to
implement, in an integrated circuit, a system inspired by the development of
multi-cellular organisms and capable of self-test and self-repair.
In Embryonics (Figure 1), an artificial organism (ORG)1 is realized by a set
of cells, distributed at the nodes of a regular two-dimensional grid. Each cell
contains a small processor coupled with a memory used to store the program
(identical for all the cells) that represents the organism’s genome. In the organ-
ism, each cell realizes a unique function, defined by a sub-program called the
gene, which is a part of the genome. The gene to be executed in a cell is se-
lected depending on the cell’s position within the organism, defined by a set of
X and Y coordinates. In Figure 1, the genes are labeled A to F for coordinates
(X, Y ) = (1, 1) to (X, Y ) = (3, 2).
The first kind of self-replication in Embryonics systems is that of the organ-
ism: an artificial organism is capable of replicating itself when there is enough
free space in the silicon circuit (at least six cells in the example of Figure 1) to
contain the new daughter organism and if the calculation of the coordinates pro-
duces a cycle. In fact, as each cell is configured with the same information (the
genome), the repetition of the vertical coordinate pattern (Y = 1 → 2 → 1 → 2)
causes the repetition of the same pattern of genes and therefore, in a sufficiently
large array, the self-replication of the organism for any number of specimens in
the X and/or the Y axes.
In Embryonics, however, there is a second replication process, which corre-
sponds to the cellular division process in biological entities, used to put in place
the initial array of cells that will then be differentiated to obtain the organism.
The need to build cells of different size and structure depending on the ap-
plication naturally led to the use of programmable logic (FPGAs) as a physical
substrate in Embryonics. Each of the computational elements of the FPGA can
then be seen as molecules, assembled in a precise configuration to form a cell.
As all cells are identical, the development process is analogous to the replication
of this configuration for as many times as there are cells in the organism.
To implement this replication, Embryonics splits the process into two phases:
– the structural phase, where a “skeleton” is created in order to divide the
physical space in a collection of groups of molecules which are empty cells;
1
Group of artificial cells that executes a given task
 Artificial Cellular Division by Self-Inspection 219

B t4 D t5 F
2 Directions of self-replication

t3

t3 t4 t5
1
A C E A

t2 DAUGHTER ORG

B t2 D t3 F t4 B t5 D
2

t1

t1 t2 t3 t4 t5
1
A C E A C E
Y
MOTHER ORG
X 1 2 3 1 2 3

Fig. 1. Self-replication of a 6-cell organism in a limited homogeneous array of 6x4 cells.

Only the expressed gene is shown in each cell.

– the configuration phase, where the configuration is sent in parallel into all
the empty cells created during the structural phase.

The structural phase is implemented by a small cellular automaton (CA) [1],

integrated in the molecular grid, capable of transforming a one-dimensional
string of states (analogous to a configuration bitstream stored in a memory
chip) into a two-dimensional structure (the blocks that will host the cells).
Figure 2 shows that the CA elements are placed in the spaces between the
molecules of the FPGA. With an appropriate sequence of states, the automaton
will be able to partition the array into identical blocks of variable size. The
“skeleton” created by the automaton during the structural phase can be seen as
the membrane of the artificial cells. Once the membrane is in place, the second
part of the cellular self-replication begins: a bitstream containing the genome
is sent to all the blocks in parallel (Figure 2), automatically creating multiple
copies of the same artificial cell. At the end of this phase, the coordinate-based
differentiation mechanism defines the structure of the organism.
This replication process is quite different from biological cellular division, as
all cells are created in parallel. Our new approach, designed to be integrated
with Embryonics, implements a cellular division much closer to reality.

3 Cell Matrix

Developed by N. Macias [4], Cell Matrix is a fine-grained reconfigurable archi-

tecture, composed of a two-dimensional grid of identical elements (referred to as
cells2 ). Each cell in the grid is interconnected with its four cardinal neighbors
and contains a lookup table (LUT) used as a truth table to define its ouputs.
2
A cell in the Cell Matix architecture can be compared to a molecule in the Embry-
onics tissue
220 E. Petraglio et al.

MEMBRANE

MEMBRANE
CFG 2 CFG 5 CFG 8 CFG 2 CFG 5 CFG 8 STATE
CONFIGURATION
PATH FPGA
ELEMENTS
CFG 1 CFG 4 CFG 7 CFG 1 CFG 4 CFG 7

CONFIGURATION
ENTRY POINTS FPGA
CONFIGURATION
INPUT SEQUENCE CFG 0 CFG 3 CFG 6 CFG 0 CFG 3 CFG 6
2112112112

Fig. 2. First, an input sequence is sent to the CA (represented by the lozenge network)
in order to set up the membrane of each cell. Then, the genome is sent in parallel to
each cell composing the ORG.

Cells of Cell Matrix circuits are capable of self-replication using a self-

inspection mechanism: each cell is at the same time configurable and able to
configure other cells without any external command. This feature is called self-
duality, and requires two different modes of operation for the cell: D-mode (Data
mode) and C-mode (Control mode). In D-mode, the cell’s LUT processes the
four input signals in order to generate output signals, whereas in C-mode, the
input data is used to fill up (configure) or re-write (re-configure) the LUT of the
cell.
The self-duality feature is the key to the self-replication mechanism of a Cell
Matrix cell. Using this mechanism, a cell is able to produce an identical copy of
itself, inspecting its own truth table and copying it to another cell in the circuit.
Figure 3 shows a 2x2 Cell Matrix grid, in which each cell has two inputs and
two outputs per side. Each cell contains an internal 16-row by 8-column truth
table, in which four 4-variable universal functions can be realized. This internal
memory governs the combinational behavior of the cell.

DNCNDNCN DNCNDNCN
DW DE DW DE
LUT
LUT

CW CE CW CE
DW DE DW DE
CW CE CW CE
DSCSDSCS DSCSDSCS

DNCNDNCN DNCNDNCN
DW DE DW DE
LUT
LUT

CW CE CW CE
DW DE DW DE
CW CE CW CE
DSCSDSCS DSCSDSCS

Fig. 3. 2x2 Cell Matrix grid. Each cell, which has two inputs and two outputs per edge,
is connected with its four direct neighbors.
 Artificial Cellular Division by Self-Inspection 221

A cell uses its D lines to exchange information with its neighbors; the nature
of this information is defined by the mode in which the cell is operating. In
D-mode, a cell reads its D input values and uses them to select one of the
16 lines of its LUT, generating the eight outputs of the cell. In C-mode, the
information sent to the cell trough the D lines is serially shifted into its LUT,
while the previous contents of the LUT are shifted out to the D outputs. Using
this mechanism, a cell is able to configure its neighbors. The C inputs are used
to define the operating mode of the cell. If any of the cell’s C inputs is 1, the
cell is in C-mode, otherwise it is in D-mode.
A cell, using its LUT, can change its C output values and therefore can
control the mode of any neighboring cell. By changing the mode of its neighbor
cells, a single cell can reach and control any cell placed in the circuit. Figure 4
shows a typical programming sequence. First, cell X configures the LUT of the Y
cell in order to transform it in a simple wire, which has to bypass C and D data
from X to Z. After this step the D and C outputs of X are directly connected
to the D and C inputs of Z, hence cell X is able to configure the LUT of Z by
sending directly its information through the D line.

(π1)
1 C 0 C 0 C 1 C
X Y Z X Y Z
π1 D 0 D π2 D D D

Fig. 4. Cell X configuring non-adjacent cell Z. On the left, cell X, by using truth table
π1, configures cell Y (its direct neighbor) as a wire. On the right, cell Y is configured
with π1 truth table and cell X can directly configure cell Z using π2 truth table.

Our approach will combine the self-replication capabilities of a Cell Matrix

architecture with the Embryonics approach in order to create a new molecule
which will better approximate the biological cellular division process in Embry-
onics systems.

4 Self-Inspection on Embryonics Tissues

In our approach, we will redefine the molecules of the Embryonics project in
a form similar to the cells of the Cell Matrix circuit (a LUT and two types
of input/output connections) in order to realize a cellular division mechanism
based on self-inspection [2] in Embryonics systems. Moreover, we will present the
design of a finite state machine (FSM), which is placed in each molecule: during
the configuration phase, this FSM will control the hardware configuration and
the communication protocol of the molecule.
By definition, in self-inspection based self-replication, a replicating system
has to generate its description by examining its own structure. Such a descrip-
tion is then used to recreate an identical copy of the original system [7]. In
222 E. Petraglio et al.

our case, the system which performs the self-replication is realized on an inte-
grated circuit which consists of a surface of silicon divided in identical elements
(i.e., molecules) distributed at the nodes of a regular two-dimensional grid. Each
molecule contains the same hardware components and only the contents of its
registers can differentiate it from its neighbors. Therefore, a self-replicating cell
creates its description by reading the contents of its register. We will say that
our system performs a simplified kind of self-inspection because it achieves self-
replication by copying the contents of its registers and not by generating the
description of the new copy.

4.1 Molecular Self-Replication

The first step of the design is the realization of a molecule capable of replicating
its content in one of its four neighbors. It has to be noted that this phase of
self-replication draws inspiration from the Cell Matrix mechanism presented in
figure 4. However, the design of our molecule is totally original as the molecule
has to fit the requirements of Embryonics.
Let us consider that the information that a molecule has to send to its neigh-
bor in order to replicate itself is fully contained in its LUT. Since the molecule,
as in the Cell Matrix cell, has one D output line per edge, it has to read and send
the content of its LUT serially. To implement this behavior, the molecule can
use its LUT as a shift-register in which input and output are connected together.
The molecule can then create a rotating bitstream and read (self-inspect) and
send out (self-replicate) its information through the D output line, as shown in
Figure 5.

DNCNDNCN

DW LUT DE
CW CE
DW DE LUT content
CW CE 1

DSCSDSCS

Fig. 5. A shift register implementing the rotating memory.

The molecule as in Cell Matrix, also has to set its C output line to 1 in
order to change the operation mode of its target neighbor. The latter is now
able to receive the information arriving form the source molecule. The end of
the replication process is signaled by a one-bit register (called overflow) that is
set when the LUT has been completely filled (the mechanism to detect the end
of the replication process in Cell Matrix is not described in the publications).
By testing the register, the target molecule knows that the replication process is
finished and sends an an acknowledgment signal through its C output line to the
 Artificial Cellular Division by Self-Inspection 223

source molecule (Figure 6). Once the configuration is over, the source molecule
returns in a quiescent state while the target molecule becomes a source molecule
and starts to self-replicate.

Acknowlegment
singal

DNCNDNCN DNCNDNCN DNCNDNCN DNCNDNCN

DW LUT DE DW LUT DE DW LUT DE DW LUT DE

CW CE CW CE CW CE CW CE
DW DE DW DE DW DE DW DE
CW CE CW CE CW CE CW CE
1
DSCSDSCS DSCSDSCS DSCSDSCS DSCSDSCS
Overflow

(a) (b)

Fig. 6. (a) First, the two molecules are in C-mode. The source molecule sends its
memory content, through the D line, into the target cell. (b) When the target molecule
is configured (i.e. the overflow bit is set), it sends back an acknowledgment signal,
which will stop the configuration process.

Waiting for CIN=1 Receiving a CIN=0 Ack=1 Ready to

configuration configuration Replicating operate

C-mode states

Fig. 7. Sequence of states describing the behavior of a molecule during the molecular
self-replication.

As a Cell Matrix cell, our molecule can operate in two different modes (C
or D). The example of Figure 6, however, defines new switching rules between
these two modes. This behavior can be represented by the state graph shown in
Figure 7.
In the Waiting for configuration state, the molecule is not configured and
quiescent. When a C input becomes 1, the molecule passes into the Receiving
configuration state, switches to C-mode, and starts filling its LUT with the
incoming configuration. When the configuration process is over, the molecule
generates an acknowledgment signal and waits for its C input value to return to
0. At this point, the molecule changes its internal state to Replicating but is still
operating in C-mode, ready to configure one of its neighbors. Once the molecule
has sent out its LUT content and received the acknowledgment signal, it returns
into D-mode and reaches the last state (Ready to operate) on the graph, where
it will remain during the normal operation of the circuit.
224 E. Petraglio et al.

4.2 Cellular Self-Replication

The next step of the design is the insertion of new hardware that will allow
the molecules to assemble in an artificial cell. It has to be noted that, in recent
publications [5,6], the Cell Matrix system has been extended to include the
possibility of grouping cells in order to compose a scructure called super-cell3 .
However, we choose to develop our own system capable of creating artificial cells.
Such a system will be expressly conceived to work with our new molecules and
will perfectly be compatible with the requirements of the Embryonics project
(self-replication and local fault-tolerance). As a consequence, the mechanisms
introduced in this section for the creation of cells are very different from those
exploited by Cell Matrix’s super-cells.
An artificial membrane, as in Embryonics cells, will be used to define which
molecules belong to which cell. This membrane will be realized with a 4-bit
register (MembReg) which will store the relative position of the molecule in the
cell (Figure 8), thus defining its state.
In a dividing cell, each molecule can be in one of the following states:

– ready to operate: the molecule has already executed its replication;

– replicating: the molecule is configuring one of its neighbors;
– ready to replicate: the molecule is waiting to start its self-replication.

Therefore, the FSM introduced above must be modified in order to handle

the new molecular states.

DNCNDNCN DNCNDNCN Memb

Reg
N S E W N S E W
DW 1 0 0 1 DE DW 1 0 1 0 DE
CW CE CW CE
LUT
LUT

DW DE DW DE
CW CE CW CE

DSCSDSCS DSCSDSCS

DNCNDNCN DNCNDNCN
N S E W N S E W
DW 0 1 0 1 DE DW 0 1 1 0 DE
CW CE CW CE
LUT
LUT

DW DE DW DE
CW CE CW CE

DSCSDSCS DSCSDSCS

Fig. 8. A cell composed of 2x2 molecules. Each MembReg is filled in order to define
the position of the molecule in the cell. N.B: If the molecule is placed in the center of
the cell the MembReg contents is equal to zero.

As an example, Figure 8 shows a set of 2x2 molecules, which will represent

the source cell that will use self-inspection to self-replicate its contents (i.e. the
3
A super-cell in the Cell Matix architecture can be compared to a cell in the Embry-
onics tissue
 Artificial Cellular Division by Self-Inspection 225

artificial genome) in order to create a daughter cell on its right. First, each
molecule in the cell will use the membrane information to know its position. In
this example, a molecule can start to replicate (replicating state) if it belongs to
the left part of the membrane; otherwise it has to wait (ready to replicate state).
The molecules in the leftmost column send out their LUT contents on the east
D output lines. The information travels through the columns of molecules in the
ready to replicate state and reaches the empty target molecules (Figure 9). The
source molecules configure the targets until they receive the acknowledgment
signals (Figure 10). At this point, the target molecules are totally configured
and in the ready to replicate state, the first column stops sending data, switches
its state from replicating to ready to operate and changes the values of its C lines
from 1 to 0. This change is picked up by the second column of molecules, which
switches its state from ready to replicate to replicating and starts to replicate.
The first cellular division is over when the second column of molecules ends its
replication.

Mother Cell

DNCNDNCN DNCNDNCN DNCNDNCN

Memb Memb Memb
DW DE DW DE DW DE
CW CE CW CE CW CE
LUT

LUT
LUT

DW DE DW DE DW DE
CW CE CW CE CW CE
1
DSCSDSCS DSCSDSCS DSCSDSCS

DNCNDNCN DNCNDNCN DNCNDNCN

Memb Memb Memb
DW DE DW DE DW DE
CW CE CW CE CW CE
LUT

LUT
LUT

DW DE DW DE DW DE
CW CE CW CE CW CE
1
DSCSDSCS DSCSDSCS DSCSDSCS

Source Column Target Column

Fig. 9. The molecules composing the source column are in the replicating state, while
the other molecules in the mother cell are in the ready to replicate state. The grey boxes
represent the configured elements, while the blank boxes represent empty elements.

The state graph for this new cellular behavior is shown in Figure 11.
It should be noted that, according to the new state graph, a cell is able to self-
replicate in one direction only. That is, the self-replication mechanism presented
above is not able to fill up an FPGA circuit, which is a two-dimensional structure.
Therefore, the molecule and the states graph have to be further modified in order
to perform a two-dimensional cellular division. A possible solution is to use a
second acknowledgment signal, Ack Cell, which will inform all the molecules in
the cell that the cellular division is over and instruct the cell has to restart its
replication in a different direction.
226 E. Petraglio et al.

Mother Cell

DNCNDNCN DNCNDNCN DNCNDNCN

Memb Memb Memb
DW DE DW DE DW DE
CW CE CW CE CW CE

LUT

LUT
LUT
DW DE DW DE DW DE
CW CE CW CE CW CE

DSCSDSCS DSCSDSCS DSCSDSCS

DNCNDNCN DNCNDNCN DNCNDNCN

Memb Memb Memb
DW DE DW DE DW DE
CW CE CW CE CW CE

LUT

LUT
LUT

DW DE DW DE DW DE
CW CE CW CE CW CE

DSCSDSCS DSCSDSCS DSCSDSCS

Source Column Target Column

(fully configured)

Fig. 10. The molecule composing the target column are totally configured. Each
molecule, using the overflow register, generates an acknowledgment signal in order
to stop the molecular self-replication.

Waiting for CIN=1 Receiving a CIN=1 Ready to CIN=0 Ack=1 Ready to

configuration configuration Ack=1 replicate Replicating operate

C-mode states

Fig. 11. Sequence of states describing the behavior of a molecule during the cellular
self-replication.

For example, if a cell is created in the bottom left corner of an FPGA, this
cell has to replicate itself to the north and then to the east. In order to fill all
the available space in the circuit, each new cell has to replicate exactly as its
mother. It should be noted that, with these growth rules, two source cells could
be configuring the same set of molecules at the same time. In order to avoid such
data collisions, only the cells on the east side of the FPGA are able to replicate
in the two dimensions. Figure 12 shows the two-dimensional growth behavior.

FPGA FPGA FPGA

Target Source

Target Source Source Target

Source Source Target Source Target

t=0 t=1 t=2

Fig. 12. Cellular division, two-dimensional growing behavior.

 Artificial Cellular Division by Self-Inspection 227

5 An Example: The LSL Organism

This section will present an extremely simplified example, the display of the
acronym “LSL”, for Logic Systems Laboratory. This acronym is considered as a
one-dimensional artificial organism composed of three cells (Figure 13a). Each
cell is located by a X coordinate, ranging from 1 to 3 in decimal or from 01 to
11 in binary. For coordinate values X = 1 and X = 3, the cell should implement
the L character, while for X = 2, it should implement the S character. A totipo-
tent cell (Figure 13b) comprises then 6 × 7 = 42 molecules, 36 of which being
invariant, five displaying the S character, and one displaying the L character. A
modulo-3 counter is embedded in the final organism and implements the truth
table shown in table 1.

Table 1. The modulo-3 counter.

Character X X1 X0 X+ X1+ X0+

L 1 0 1 2 1 0
S 2 1 0 3 1 1
L 3 1 1 1 0 1

The modulo-3 counter is represented by the logic diagram and symbol of

figure 15. According to the table, the value of the binary variable X0 is sufficient
to distinguish the display of character L (X0 = 1) from the display of character
S (X0 = 0 or X0 = 1). It is important to note that the modulo-3 counter is
designed to produce a cycle in the calculation of the coordinates, in our example:
X = 1 → 2 → 3 → 1... and Y = 1 → 1 → 1.... This coordinate calculation will
differentiate the cells in order to produce as artificial organisms composed of
three cells as it is possible (Figure 14).

S S

S S
L S

X= 1 2 3
(a) (b)

Fig. 13. (a) The three cells of the artificial organism. (b) The totipotent cell.

These specifications are sufficient to design the final architecture of the

totipotent cell, which is shown in figure 16. The final cell is composed of seven
different types of molecules, each of which realizes one of the following tasks:
228 E. Petraglio et al.

Artificial Organism

Y=1

X=1 2 3 1 2 3

Fig. 14. A population of artificial organisms obtained by the repetition of the cellular
coordinate X = 1 → 2 → 3 → 1... and Y = 1 → 1 → 1....

X0 X0+
X0 X0+
X1
INC X1+
X1 X1+

Fig. 15. Logic diagram ans symbol of the incrementer.

– 0: Function-less molecule.
– 1: Two horizontal busses, which carry X0 and X1 signals.
– 2: Modulo 3 intcrementation.
– 3: One vertical bus, which carries the X0 signal.
– 4: Permanent switch-on display.
– 5: Display of S character only (X0 = 0 or X0 = 1).
– 6: Display of L character only (X0 = 1).

Figure 16 shows that each molecule composing the totipotent cell has to
memorize two characters. The first character, which is in brackets, represents
the hexadecimal value memorized in the MembReg and used to construct the
artificial membrane (Table 2), while the second character, which is in bold font,
controls the molecular behavior and can be coded using three bits. Therefore, in
order to implement this example, the logic core of our new molecule is composed
by a 4×1 LUT, which is used to memorize the second character controlling the
task that the molecule has to realize.
The artificial genome, which has to be injected in the FPGA in order to
create the first totipotent cell is represented in figure 17. It has to be noted that
each row represents the configuration stream of a row of molecules, and can be
splitted in several groups of 8-bits each group representing the configuration of
 Artificial Cellular Division by Self-Inspection 229

(9) 1 (8) 1 (8) 1 (8) 1 (8) 1 (A) 2

X0 X0+
X1
INC X1+

(1) 0 (0) 4 (0) 5 (0) 5 (0) 0 (2) 0

(1) 0 (0) 4 (0) 3 (0) 3 (0) 0 (2) 0

(1) 0 (0) 4 (0) 5 (0) 5 (0) 0 (2) 0

(1) 0 (0) 6 (0) 3 (0) 5 (0) 0 (2) 0

(1) 0 (0) 4 (0) 4 (0) 4 (0) 0 (2) 0

(5) 0 (4) 0 (4) 0 (4) 0 (4) 0 (6) 0

Fig. 16. The final totipotent cell made up of 6×7 = 42 molecules

a single molecule (i.e. membrane information + molecular task). Analyzing the

membrane information of the first row we can see that this flow of information
will configure the south row of the cell.
Last, it was possible to embed the basic molecule in each of the 2000 field-
programmable gate arrays of the BioWall [12] and to show the growth of the
the multicellular artificial organism, followed by its selfreplication in both verti-
cal and horizontal dimensions (Figure 18). We therefore, obtain a population of
identical organisms (i.e. clones) thus creating a fourth level in our hierarchy, a
population of organisms.
In figure 18 it is interesting to note that two cells are displaying a corrupted
line (or column). This is the characteristic behavior of a cell during its replication
phase. In fact, a replicating cell, inspects its content in order to sent it to its
daughter. Therefore, during this phase, the cell is not able to correctly perform
its task (in our example: display a L or a S).
230 E. Petraglio et al.

Table 2. Possible contents of the MembReg register

HEX value N S E W position in the cell

0 0 0 0 0 molecule placed in the center of the cell
1 0 0 0 1 west molecule
2 0 0 1 0 east molecule
4 0 1 0 0 south molecule
5 0 1 0 1 southwest molecule
6 0 1 1 0 southeast molecule
8 1 0 0 0 north molecule
9 1 0 0 1 northwest molecule
A 1 0 1 0 northeast molecule
others - - - - invalid combination

Fig. 17. The artificial genome of the totipotent cell.

6 Conclusion

This paper presented a novel approach for self-replication for Embryonics sys-
tems, closer to biological cellular division than past mechanisms. The new
molecule thus introduced will be the basic block of a new generation of Em-
bryonics circuits, and is designed to be integrated into a more complex system
to implement a full developmental (growth) process in digital hardware.
However, self-replication, as described, is not a very useful feature for current
real-world applications, as self-replication is completely deterministic. In other
words, its result is a predictable FPGA configuration, which could be realized
by conventional FPGA place and route tools.
This argument, however, is not valid if the self-replication process embeds
fault-detection and self-repair mechanisms to, for example, grow artificial or-
ganisms on an FPGA with fabrication flaws. In fact, standard FPGA place and
route tools are not capable of generating bitstream configuration files for FPGA
circuits which are not fault-free. A cell composed of molecules capable of au-
 Artificial Cellular Division by Self-Inspection 231

Fig. 18. The BioWall implementation

tonomous self-replication, fault-detection and self-repair, however, will be able

to handle flawed FPGAs by adapting its structure so as to avoid faulty areas
of the circuit. The next step in the design of the new Embryonics molecule will
address this issue and realize autonomous self-replication in faulty circuits.
Moreover, the predicted development of molecular-level electronics (such as
nanotechnologies) implies a forthcoming need for mechanisms that will allow a
circuit to structure itself, rather than having a structure imposed at fabrication.
A cellular division approach such as the one described in this article could be a
useful tool to achieve this kind of self-organization.

References
1. Codd, E., F.: Cellular Automata. Academic Press, New York (1968)
2. Ibáñez, J., Anabitarte, D., Azpeitia, I., Barrera, O., Barrutieta, A., Blanco, H., and
Echarte, F. Self-inspection based reproduction in cellular automata. Proc. 3rd Eur.
Conf. on Artificial Life (ECAL95), LNCS 929, Springer, Berlin (1995) 564–576
3. Jenkins, J.: Designing with FPGAs and CPLDs. Prentice Hall, Englewood Cliffs,
NJ (1994)
4. Macias, N.: The PIG Paradigm: The Design and Use of a Massively Paral-
lel Fine Grained Self-Reconfigurable Infinitely Scalable Architecture. Proc. 1st
NASA/DOD Workshop on Evolvable Hardware (EH1999), IEEE Comp. Soc.
(1999) 175–180
232 E. Petraglio et al.

5. Durbeck, L., Macias N.: Defect-tolerant, fine-grained parallel testing of a Cell Ma-
trix. Proc. SPIE ITCom 2002, Series 4867, (2002) 71–85
6. Durbeck, L., Macias N.: Self-Assembling Circuits with Autonomous Fault Han-
dling. Proc. 4th NASA/DOD Conference on Evolvable Hardware (EH’02), IEEE
Comp. Soc. (1999) 46–55
7. Laing R.: Automaton Models of Reproduction by Self-inspection. Journal of The-
oretical Biology, volume 66, (1977) 437-456
8. Mange, D., Sipper, M., Stauffer A., Tempesti, G.: Towards Robust Integrated Cir-
cuits: The Embryonics Approach. Proceedings of the IEEE 88 (2000) 516–541
9. Sanchez, E.: Field Programmable Gate Array (FPGA) Circuits. Towards Evolvable
Hardware, LNCS 1062, Springer-Verlag, Heidelberg (1996) 1–18
10. Tempesti, G., Mange, D., Stauffer A.: The Embryonics Project: a Machine Made
of Artificial Cells. Rivista di Biologia-Biology Forum 92 (1999) 143–188
11. Trimberger, S.: Field-Programmable Gate Array Technology. Kluwer Academic,
Boston, MA (1994)
12. Tempesti G., Mange D., Stauffer A., Teuscher C.: The BioWall: an Electronic
Tissue for Prototyping Bio-Inspired Systems. Proc. 4th NASA/DOD Conference
on Evolvable Hardware (EH’02), IEEE Comp. Soc. (2002) 221–230
Another Random Scribd Document
with Unrelated Content
back
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookball.com