Cover page

Image Name: Cyber Security: Crime and cheating on the web is increasing. One should be prepared. Image

Source: AI generated image by Jinsu Ann Mathew

Managing Editor Chief Editor Editorial Board Correspondence

Ninan Sajeeth Philip Abraham Mulamoottil K Babu Joseph The Chief Editor

Ajit K Kembhavi airis4D

Geetha Paul Thelliyoor - 689544

Arun Kumar Aniyan India

Sindhu G

Journal Publisher Details

Publisher : airis4D, Thelliyoor 689544, India

Website : www.airis4d.com

Email : nsp@airis4d.com

Phone : +919497552476

Editorial

by Fr Dr Abraham Mulamoottil

airis4D, Vol.3, No.2, 2025

www.airis4d.com

This issue of airis4D highlights the release of

DeepSeek in January 2025. Its open-source approach

and groundbreaking advancements set new benchmarks

for AI innovation. Unlike many proprietary models,

DeepSeek champions transparency, collaboration, and

cost-eﬀective performance. It rivals top AI models

at a fraction of the cost, making advanced AI

more accessible and energy-eﬃcient. It reduces

operational costs and environmental impact while

achieving signiﬁcant improvements in inference speed

and enhancing user experience.

The article of Blesson George on “Multimodal

Transformers” extends traditional transformer models

beyond text to process multiple data types—such as

images, audio, and video—enabling a more human-

like understanding of information. These models

use self-attention and cross-modal fusion to integrate

diverse inputs eﬀectively. Key applications include

medical diagnosis, behaviour analysis, vision-language

tasks, and recommendation systems. Diﬀerent fusion

techniques, such as early fusion, late fusion, and

cross-attention fusion, optimize how data is combined

and interpreted. Multimodal representation learning

remains a challenge, requiring smooth, structured,

and robust embeddings. Despite ongoing research,

multimodal transformers are revolutionizing AI by

making it more versatile, interpretable, and adaptable

across industries.

The article X-ray Astronomy: Through Missions

by Aromal P, explores key X-ray astronomy

missions from the 1990s that signiﬁcantly advanced

our understanding of high-energy astrophysical

phenomena. It highlights NASA’s Rossi X-ray Timing

Explorer (RXTE), launched in 1995, which provided

groundbreaking insights into neutron stars, black holes,

and gamma-ray bursts over its 16-year operational

period. The article also discusses India’s IRS-P3

satellite, which, despite being a remote sensing satellite,

carried the Indian X-ray Astronomy Experiment

(IXAE), contributing valuable data on X-ray binaries.

Another major mission covered is Italy’s BeppoSAX,

launched in 1996, which played a pivotal role in the

study of gamma-ray burst afterglows and high-energy

astrophysical sources. The article emphasizes how data

from these missions continue to shape X-ray astronomy

research even decades after their decommissioning.

T Corona Borealis (T CrB) by Sindhu G is a

fascinating binary star system located about 1,300

light-years away in the Corona Borealis constellation.

Comprising a white dwarf and a red giant, it is

classiﬁed as a cataclysmic variable due to its periodic

outbursts caused by mass transfer between the two stars.

First documented in 1865, T CrB undergoes dramatic

increases in brightness as the white dwarf accretes

material from its red giant companion, forming an

accretion disk that periodically ignites. With an orbital

period of approximately 31.6 days, this system provides

key insights into stellar evolution, binary interactions,

and mass transfer processes.

Observations across multiple wavelength including

optical, infrared, and X-ray help astronomers track these

outbursts and improve predictive models. Scientists

are currently monitoring T CrB in anticipation of its

next Nova explosion, expected around 2025. This study

highlights the system’s signiﬁcance in understanding the

lifecycle of stars, with ongoing research oﬀering deeper

insights into the dynamics of cataclysmic variables.

The article “Synthetic Biology – A Revolutionary

Scientiﬁc Frontier by Geetha Paul highlights that

synthetic biology is a transformative ﬁeld that applies

engineering principles to biological systems, treating

them as programmable entities. Unlike traditional

genetic engineering, it seeks to create entirely new

biological components and organisms with precise

control. Researchers use advanced tools like CRISPR,

DNA sequencing, and AI to manipulate genetic

material, leading to groundbreaking applications in

medicine, environmental sustainability, and industry.

Key innovations include engineered cells for targeted

therapies, microorganisms that degrade plastic waste,

and synthetic biological circuits. Ethical considerations

and regulatory frameworks are crucial to ensure

responsible development. As technology advances,

synthetic biology holds immense potential to address

global challenges in healthcare, conservation, and

industrial production.

The article ”Domain Generation Algorithm (DGA)

Detection Techniques” by Jinsu Ann Mathew explores

modern techniques for detecting domains generated

by Domain Generation Algorithms (DGAs), which

cybercriminals use to evade traditional security

measures. As DGAs become more advanced, detection

strategies have evolved to counter them. Key detection

methods include: Statistical Analysis: Which identiﬁes

anomalies in character distribution, domain length, and

entropy to ﬂag suspicious domains. Machine Learning

(ML): Uses supervised, unsupervised, and deep learning

models to classify domains based on learned patterns.

N-gram Analysis: Examines character sequences

to diﬀerentiate human-created domains from DGA-

generated ones. DNS Query Patterns: Monitors query

frequencies, time-based behaviours, and newly observed

domains to detect malicious activity. By integrating

these techniques, security systems can build a robust,

adaptive defence against evolving DGA threats. The

article highlights the importance of interdisciplinary

research in strengthening cybersecurity measures.

iii

News Desk

Marry Gormally from USA visited airis4D on January 29th 2025.

Contents

Editorial ii

I Artiﬁcial Intelligence and Machine Learning 1

1 Multimodal Transformers 2

1.1 Types of Modalities in Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Diﬀerent Fusion Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

II Astronomy and Astrophysics 5

1 X-ray Astronomy: Through Missions 6

1.1 Satellites in 1990s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 T Corona Borealis: A Comprehensive Study of a Binary Star System 10

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2 Discovery and Historical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.3 The Components of T Corona Borealis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.4 Cataclysmic Variability and Outbursts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.5 Orbital Characteristics and Periodicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.6 The Role of T Corona Borealis in Stellar Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.7 Observations and Research on T Corona Borealis . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.8 Future Prospects and Challenges in Studying T Corona Borealis . . . . . . . . . . . . . . . . . . 12

2.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

III Biosciences 14

1 Synthetic Biology: A Revolutionary Scientiﬁc Frontier 15

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

1.2 Embedded Hybridisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.3 The Fundamental Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.4 Technological and Healthcare Innovations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.5 GES (General Environmental Solutions) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.6 Environmental and Industrial Solutions (EIS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.7 Ethical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.8 Future Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

CONTENTS

IV General 20

1 Domain Generation Algorithm (DGA) Detection Techniques 21

1.1 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

1.2 Machine Learning (ML) Approaches: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1.3 N-gram Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1.4 DNS Query Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Part I

Artiﬁcial Intelligence and Machine Learning

Multimodal Transformers

by Blesson George

airis4D, Vol.3, No.2, 2025

www.airis4d.com

In the rapidly evolving ﬁeld of artiﬁcial

intelligence, transformers have revolutionized the way

machines process and understand data. Originally

designed for natural language processing, transformers

have now been extended beyond text-based tasks to

incorporate multiple modalities—such as images, audio,

and video—giving rise to multimodal transformers.

These advanced AI models are capable of processing

and integrating information from diﬀerent data sources

to perform more complex and human-like understanding

of the world.

1.0.1 What Are Multimodal Transformers?

Multimodal transformers are a class of AI models

that can simultaneously process and relate multiple

types of input data. Unlike traditional unimodal models

that operate on a single type of data (e.g., text-only or

image-only models), multimodal transformers leverage

self-attention mechanisms to combine textual, visual,

and auditory information eﬀectively. This capability

allows them to generate richer, more context-aware

predictions and responses.

1.0.2 How Do They Work?

At their core, multimodal transformers use a

shared embedding space where diﬀerent data types

are transformed into a common representation. These

models employ cross-modal attention mechanisms to

allow interactions between various inputs, enhancing

the model’s ability to understand relationships between

diﬀerent modalities. For example, a multimodal

transformer that analyzes an image with a caption will

align relevant words with the corresponding visual

elements to provide a coherent interpretation.

1.1 Types of Modalities in Machine

Learning

Multimodal learning involves various types of

data modalities, which can be categorized into four

main groups. Tabular data is represented as structured

datasets where observations are rows and features are

columns. Graphs are datasets where observations

are represented as vertices, with edges capturing

relationships between them. Signals include data in ﬁle

formats such as images (.jpeg), audio (.wav), and other

structured numerical representations. Sequences are

structured data such as text, where features correspond

to diﬀerent elements in the sequence, including

characters, words, or documents.

In real-world applications, multimodal models

incorporate various data sources. Labels are an example

of tabular data, while reviews, titles, and descriptions

represent textual data. Images of products and movie

posters are examples of visual data, and relations

between movies, such as those seen by one user, are

represented as graphs. These diverse data sources

allow multimodal models to gain a comprehensive

understanding of information.

1.1.1 Examples of Multimodal Tasks

Multimodal learning has been successfully applied

in various domains. In medical diagnosis, multimodal

AI integrates medical images, patient demographics,

1.2 Diﬀerent Fusion Techniques

and genetic data to predict diseases at an early

stage. Behavior analysis utilizes audio, video, and

physiological signals to assess human emotions and

stress levels. Vision-language tasks involve models

that perform question answering based on textual and

image data, such as identifying objects in an image

or explaining their signiﬁcance. Recommendation

systems enhance user experiences by combining text

descriptions, images, and social graphs to suggest

personalized content, such as movies or fashion

recommendations.

1.1.2 Multimodal Representation Learning

Eﬀective multimodal representation learning

involves capturing meaningful embeddings from

multiple modalities. While unimodal models like BERT

(text) and ResNet (images) have been successful in their

respective domains, a universal approach for integrating

multimodal data is still being developed. A good

multimodal representation should ensure smoothness,

meaning that similar inputs should produce similar

embeddings. It should maintain manifold structures,

where data points belonging to similar categories cluster

together. Natural clustering should allow categorical

labels to be assigned based on common features, and

sparsity should ensure that representations focus on the

most relevant features while avoiding redundancy.

Additionally, eﬀective multimodal representation

should preserve similarities between diﬀerent

modalities in a joint embedding. It should also be robust

to missing modalities, allowing meaningful inference

even when some data sources are unavailable.

1.2 Diﬀerent Fusion Techniques

Multimodal transformers use diﬀerent fusion

techniques to integrate multiple modalities eﬀectively:

1.2.1 Early Fusion (Joint Embedding Space)

In this approach, diﬀerent modalities (e.g., text and

image) are combined at the input stage and mapped into

a shared embedding space. This allows the model to

learn joint representations early in the process, capturing

relationships between modalities from the beginning.

This method is eﬃcient in ensuring tight integration

between diﬀerent types of input but requires large,

well-annotated datasets to work eﬀectively.

Figure 1: Late fusion architecture

1.2.2 Late Fusion (Separate Encodings)

Here, each modality is processed separately

through independent encoders, and their representations

are combined only at the ﬁnal decision-making stage.

This method is useful when individual modalities

need to be analyzed independently before integrating

their outputs. Late fusion is often more interpretable

than early fusion since decisions can be traced

back to speciﬁc modalities. It is widely used in

applications where each modality contributes distinct

and complementary information, such as combining

audio and text in sentiment analysis.

1.2.3 Cross-Attention Fusion

This technique enables direct interactions between

diﬀerent modalities by using cross-attention layers,

where features from one modality inﬂuence the

representation of another. Instead of simply

concatenating features, cross-attention allows the model

to dynamically adjust its understanding of one modality

based on the other. This is particularly useful in

tasks like image captioning and video analysis, where

the model must understand which words align with

which visual components. Models like CLIP use

cross-attention mechanisms to create powerful vision-

language representations.

1.3 Conclusion

Multimodal transformers represent a

groundbreaking advancement in artiﬁcial intelligence,

pushing the boundaries of machine understanding by

integrating multiple data modalities. By combining

1.3 Conclusion

text, images, audio, and other forms of information,

these models oﬀer a more comprehensive and

accurate interpretation of complex data. Their

applications across various ﬁelds, including healthcare,

human-computer interaction, and content generation,

demonstrate their transformative potential. As research

continues, the development of more eﬃcient and

interpretable multimodal systems will further enhance

AI’s capabilities, making it more responsive and

adaptable to real-world challenges. The future of AI

lies in harnessing the power of multimodal transformers

to create smarter, more versatile, and human-like

intelligent systems.

References

Multimodal Models and Fusion - A Complete

Guide

Eﬀective Techniques for Multimodal Data Fusion:

A Comparative Analysis

About the Author

Dr. Blesson George presently serves as

an Assistant Professor of Physics at CMS College

Kottayam, Kerala. His research pursuits encompass

the development of machine learning algorithms, along

with the utilization of machine learning techniques

across diverse domains.

Part II

Astronomy and Astrophysics

X-ray Astronomy: Through Missions

by Aromal P

airis4D, Vol.3, No.2, 2025

www.airis4d.com

1.1 Satellites in 1990s

So far, we have discussed the X-ray astronomical

satellites launched before 1995. In the previous article,

we mentioned that during the 1990s, the golden age

of astronomy had begun. Now, we will discuss more

about the satellites that contributed extensively to our

knowledge about X-ray emitting sources.

1.1.1 RXTE

The Rossi X-ray Timing Explorer (RXTE)

developed, launched, and operated by NASA was one of

the signiﬁcant leaps toward X-ray astronomy. It had the

best timing resolution for an X-ray astronomical satellite

launched until then, and its timing capabilities matches

with any other instruments launch in this decade as

well. This satellite was launched from Kennedy Space

Center on December 30, 1995, using a Delta II rocket.

It was then placed in a low earth circular orbit of radius

580 km and inclination of

○

, which had an orbital

period of 90 minutes. Initially planned for 2 years,

RXTE stayed operational for 16 years till January 5,

2012, providing data that changed our views about the

neutron stars and black holes. This satellite had covered

an energy range of 2-250 keV.

Three scientiﬁc payloads were boarded in RXTE:

Proportional Counter Array (PCA): PCA

consisted of ﬁve identical Xenon-based

proportional counters, each with an eﬀective

area of

1300 cm

and operated in the 2-60 keV

energy range. It had a total eﬀective area of 6500

and observed the sky with an FOV of

○

and

a timing resolution of 1

seconds. PCAs were

Figure 1: Cut-away view of XTE Satellite.

Credit: C.A. Glasser et al. 1994

designed, built, and tested within the Goddard

Space Flight Center (GSFC) Laboratory for High

Energy Astrophysics.

High Energy X-ray Timing Experiment

(HEXTE): HEXTE consisted of eight NaI/CsI

scintillation detectors optically coupled to a

photomultiplier tube. A lead honeycomb

collimated the detectors, and they were

interdependently grouped into two clusters, each

with their own data processing systems. With

a net eﬀective area of 1600

, the detectors

worked in the 20-200 keV energy range with a

timing resolution of 8

seconds. This instrument

was developed by the University of California at

San Diego.

All-Sky Monitor (ASM): ASM consisted of three

position-sensitive proportional cameras mounted

over a rotating pedestal. While one pointed along

the axis of rotation of RXTE, the other two rotated

perpendicular to it. The total eﬀective area of

ASM was 90

and worked in 2-10 keV energy.

It scanned 80% of the sky in 90 minutes and had

1.1 Satellites in 1990s

a FOV of 6

○

× 90

○

RXTE was an extremely successful mission, which

discovered many interesting phenomena related to

compact objects, including the kilohertz Quasi-Periodic

Oscillations (kHz QPO) and burst oscillations in

Neutron Star systems, as well as the high-frequency

oscillations in the Black Hole systems. It also

discovered the ﬁrst accreting millisecond pulsar, which

proved how the radio-emitting millisecond pulsars are

formed. Later, it discovered many more accreting

millisecond pulsars. Earlier, it was believed that the

accreting millisecond pulsars did not show any burst

phenomenon due to the magnetic ﬁeld conﬁnement.

However, RXTE observed the burst phenomena from

an accreting millisecond X-ray pulsar for the ﬁrst time.

RXTE also detected the afterglow of a gamma-ray burst.

Even after a decade of its decommission, RXTE data

continues to be analyzed to bring new science scenarios.

1.1.2 IRS-P3

Indian Remote Sensing Satellite-P3 was launched

on March 21, 1996, by Polar Satellite Launch

Vehicle(PSLV) into an 817 km circular orbit at

an inclination of 98.7

○

. Even though it was a

remote sensing satellite, it carried an experimental

X-ray astronomy instrument called the Indian X-

ray Astronomy Experiment (IXAE). IXAE payload

consisted of three identical pointed proportional

counters (PPC) and one X-ray Sky Monitor (XSM). PPC

worked in an energy range of 2-20 keV, and XSM worked

in a 2-8 keV energy range. During its operation, IXAE

provided valuable information about X-ray binaries.

1.1.3 BEPPOSAX

”Satellite per Astronomia X” (SAX, which means

Satellite for X-ray Astronomy)was an Italian satellite

developed to study high-energetic phenomena. After

the launch, the satellite was renamed in honor of

Italian physicist Giuseppe Occhialini( Giuseppe Paolo

Stanislao ”Beppo” Occhialini) as BeppoSAX. The

Mission was supported by a consortium of Italian

institutes, institutes in the Netherlands, and ESA’s space

science department. BeppoSAX was launched on April

Figure 2: BeppoSAX scientiﬁc payload

accommodation

Credit: G. Boella et al. 1997

30, 1996, into a circular orbit with a radius of 600

km by Atlas-Centaur rocket. At an inclination of

3.9

○

and an orbital period of 96 minutes, the orbit provided

a safe escape from the South Atlantic Anomaly. It

used the screening of Earth’s magnetism to reduce

cosmic wave background. The satellite operated for 6

years and was decommissioned in April 2002. Using

diﬀerent instruments, BeppoSAX studied high energetic

phenomena in an energy range of 0.1-300 keV. There

were 5 scientiﬁc instruments in BeppoSAX:

The Low Energy Concentrator Spectrometer

(LECS): An imaging gas scintillation

proportional counter sensitive to X-rays

in the energy range of 0.1-10 keV. It had an

eﬀective area of 124

and a time resolution

of 16µs.

The Medium Energy Concentrator Spectrometers

(MECS): It consisted of a grazing incidence

Mirror Unit and a position-sensitive Gas

Scintillation Proportional Counter. BeppoSAX

had three identical units of MECS that worked

in an energy range of 1.3-10 keV, with each unit

having an eﬀective area of 124 cm

High-Pressure Gas Scintillation Proportional

Counter (HPGSPC): A Xenon-based proportional

counter(Xenon-90% Helium-10%) with an

eﬀective area of 450

. It operated in the

energy range of 3-120 keV.

Phoswich Detector System (PDS): It consisted

of a square array of four independent

NaI(Tl)/CsI(Na) PHOsphor sandWICH

(PHOSWICH) scintillation detectors. Each of

the four detectors was made of two crystals

1.1 Satellites in 1990s

of NaI(Tl) and CsI(Na), optically coupled and

forming what is known as PHOSWICH. PDS

had an eﬀective area of 640

and worked in

the 15-300 keV energy range.

Wide Field Camera(WFC): A coded mask

aperture developed by Space Research

Organization Netherlands(SRON). BeppoSAX

had two WFCs onboard, which worked in the

energy range of 2-30 keV, suitable for imaging

high-energy X-rays. It had an eﬀective area of

600 cm

Among all instruments, LECS, MECS, HPGSPC,

and PDS were narrow-ﬁeld instruments. One of

the main features of BeppoSAX was its broadband

coverage, starting from 0.1 keV to 300 keV. BeppoSAX

studied numerous galactic and extragalactic sources and

discovered new sources as well. It studied the afterglow

after a gamma-ray burst for the ﬁrst time along with

RXTE.

References

Santangelo, Andrea and Madonia, Rosalia and

Piraino, Santina A Chronological History of X-

Ray Astronomy Missions.Handbook of X-ray and

Gamma-ray Astrophysics.ISBN 9789811645440

RXTE Image Gallery

Proportional Counter Array (PCA)

Keith Jahoda, Craig B. Markwardt, Yana Radeva,

Arnold H. Rots, Michael J. Stark, Jean H.

Swank, Tod E. Strohmayer, and William Zhang

Calibration of the Rossi X-ray Timing Explorer

Proportional Counter Array

C.A. Glasser, C.E. Odell, and S.E. Seufert he

Proportional Counter Array (PCA) Instrument

for the X-ray Timing Explorer Satellite

(XTE) IEEE TRANSACTIONS ON NUCLEAR

SCIENCE, VOL. 41, NO. 4, AUGUST 1994

R. E. Rothschild, P. R. Blanco, D. E. Gruber, W.

A. Heindl, D. R. MacDonald, D. C. Marsden, M.

R. Pelling, L. R. Wayne, and P. L. Hink In-Flight

Performance of the High Energy X-Ray Timing

Experiment on the Rossi X-Ray Timing Explorer

ApJ 496 538

Parmar, A. N, Martin, D. D. E. , Bavdaz, M. ,

Favata, F, Kuulkers, E. , Vacanti, G., Lammers,

U, Peacock, A. and Taylor, B. G. The low-

energy concentrator spectrometer on-board the

BeppoSAX X-ray astronomy satellite A & A

Supplement series, Vol. 122, April II 1997,

p.309-326.

G. Boella1, L. Chiappetti, G. Conti, G.

Cusumano, S. Del Sordo, G. La Rosa, M.C.

Maccarone, T. Mineo, S. Molendi, S. Re,

B. Sacco, and M. Tripiciano The medium-

energy concentrator spectrometer on board the

BeppoSAX X-ray astronomy satellite A & A

Supplement series, Vol. 122, April II 1997,

p.327-340.

Instrument description

The BeppoSAX Phoswich Detector System

(PDS)

R. Jager, W.A. Mels, A.C. Brinkman, M.Y.

Galama, H. Goulooze, J. Heise, P. Lowes,

J.M. Muller, A. Naber, A. Rook, R. Schuurhof,

J.J. Schuurmans, and G. Wiersma The Wide

Field Cameras onboard the BeppoSAX X-ray

Astronomy Satellite A & A Supplement series,

Vol. 125, November I 1997, p.557-572.

G. Boella, R.C. Butler, G.C. Perola, L. Piro,

L. Scarsi, and J.A.M. Bleeker BeppoSAX, the

wide band mission for X-ray astronomy A &

A Supplement series, Vol. 122, April II 1997,

p.299-307.

Indian X-ray Astronomy Experiment (IXAE)

Indian Remote Sensing Satellite IRS-P3

1.1 Satellites in 1990s

About the Author

Aromal P is a research scholar in

Department of Astronomy, Astrophysics and Space

Engineering (DAASE) in Indian Institute of

Technology Indore. His research mainly focuses on

studies of Thermonuclear X-ray Bursts on Neutron star

surface and its interaction with the Accretion disk and

Corona.

T Corona Borealis: A Comprehensive Study

of a Binary Star System

by Sindhu G

airis4D, Vol.3, No.2, 2025

www.airis4d.com

Figure 1: Corona Borealis. (Image Credit: seasky.org)

2.1 Introduction

T Corona Borealis (T CrB) is one of the most

intriguing binary star systems in the sky. Situated

approximately 1,300 light-years away from Earth in the

constellation of Corona Borealis (Figure: 1 and Figure:

2), T CrB (Figure: 3)has attracted considerable attention

in the astronomical community. Its combination of

diﬀerent stellar types—especially its characteristics as a

cataclysmic variable star—makes it a fascinating subject

for both amateur and professional astronomers alike.

In this article, we will explore the star system’s

discovery, its key features, the physical characteristics

of its components, its behavior over time, and its

signiﬁcance to the ﬁeld of stellar evolution and

astronomical research.

Figure 2: IAU Corona Borealis chart. (Image Credit:

IAU and Sky and Telescope magazine (Roger Sinnott

and Rick Fienberg))

Figure 3: Chart showing location of T Coronae Borealis

in constellation of Corona Borealis. (Image Credit:

DE/The Guardian)

2.2 Discovery and Historical Background

2.2 Discovery and Historical

Background

The star system T Corona Borealis was ﬁrst

cataloged in 1865, when astronomers discovered its

unusual variability. The system’s designation, ”T CrB,”

comes from its placement in the constellation Corona

Borealis, the Northern Crown. It is a binary system,

composed of a white dwarf and a red giant star, and

has been extensively studied due to its variability and

unique features.

The most remarkable characteristic of T CrB is its

classiﬁcation as a cataclysmic variable. This means that

it experiences periodic outbursts, which are associated

with signiﬁcant increases in luminosity. These outbursts,

though not as dramatic as those in some other variable

stars, are indicative of the complex interaction between

the two stars in the binary system.

2.3 The Components of T Corona

Borealis

2.3.1 The White Dwarf

At the heart of the T Corona Borealis system lies

a white dwarf star. White dwarfs are the remnants of

stars that were once similar in size to our Sun but have

exhausted their nuclear fuel and shed their outer layers.

What remains is an incredibly dense core, which is the

white dwarf.

The white dwarf in T CrB is primarily composed

of carbon and oxygen, with a mass roughly 1.2 times

that of the Sun, but with a volume similar to Earth. Its

surface temperature is high, ranging from 20,000 to

30,000 K, which makes it very hot compared to cooler

stars, such as the red giant companion.

2.3.2 The Red Giant Companion

The companion star to the white dwarf in T CrB

is a red giant. Red giants are stars that are in a later

phase of their evolution, having expanded and cooled

after exhausting their core hydrogen supply. The red

giant in T CrB is roughly 3 times the mass of the Sun

Figure 4: Artist’s impression of a white dwarf drawing

material away from its red giant partner. (Image Credit:

NASA/CXC/M.Weiss)

and has a much larger radius, which extends far beyond

the size of the white dwarf.

The interaction between these two stars—one a

degenerate, dense remnant and the other a bloated,

expanding giant—forms the basis of the cataclysmic

variable behavior. The red giant transfers material via

its outer layers to the white dwarf, which accretes matter

and exhibits dramatic increases in brightness during

certain intervals(Figure: 4).

2.4 Cataclysmic Variability and

Outbursts

The term cataclysmic variable refers to stars that

experience dramatic ﬂuctuations in brightness over

time. In the case of T Corona Borealis, this variability

is driven by the mass transfer between the two stars.

The red giant is losing material to the white dwarf via

an accretion disk, and this process produces periodic

bursts of energy.

The mass transfer from the red giant creates a disk

of hot, ionized gas around the white dwarf. As this gas

spirals inward, it heats up and emits radiation, causing

the system’s brightness to ﬂuctuate. These outbursts

typically occur on a timescale of decades, but the system

remains in a constant state of mass exchange.

2.5 Orbital Characteristics and Periodicity

2.5 Orbital Characteristics and

Periodicity

T Corona Borealis has an orbital period of about

31.6 days, meaning that the two stars in the system

orbit each other roughly once a month. The separation

between the two stars is around 0.15 AU (astronomical

units), which is quite close in terms of stellar systems.

Due to the proximity of the stars, gravitational

interactions play a signiﬁcant role in shaping the

system’s behavior. The white dwarf pulls material

from the outer layers of the red giant, and this mass

transfer can cause temporary changes in the system’s

luminosity, as mentioned earlier. The orbital period

is important for astronomers because it helps them

determine the system’s mass and size, which provides

insight into the properties of both stars.

2.6 The Role of T Corona Borealis in

Stellar Evolution

The study of T Corona Borealis provides valuable

insights into the evolution of binary star systems. As

a cataclysmic variable, it is a key example of how

mass transfer between binary stars can inﬂuence their

development. In particular, the white dwarf in this

system is a perfect example of a star that has undergone

signiﬁcant mass loss and transformation after exhausting

its fuel.

Understanding such systems also aids astronomers

in studying the fate of stars similar to the Sun. As stars

like the Sun age, they will eventually become red giants,

shedding material onto companion stars. In some cases,

this will lead to the formation of white dwarfs and other

exotic remnants.

2.7 Observations and Research on T

Corona Borealis

Astronomers use a variety of techniques to study

T Corona Borealis, from ground-based telescopes to

space-based observatories. Observations at diﬀerent

wavelengths—including optical, infrared, and X-

ray—have provided insights into the complex dynamics

of the system.

For example, in X-rays, the accretion disk around

the white dwarf becomes a powerful emitter of high-

energy radiation. These emissions are often studied

using space-based telescopes like the Chandra X-ray

Observatory and the XMM-Newton satellite.

In addition, optical observations have been used

to track the periodic outbursts and better understand

the mass transfer process. Through careful monitoring,

astronomers can measure the light curves and predict

the timing of the next outbursts, which are crucial for

advancing our understanding of cataclysmic variables.

2.8 Future Prospects and Challenges

in Studying T Corona Borealis

While much has been learned about T Corona

Borealis, there is still much to discover. The complexity

of the mass transfer and accretion processes is not fully

understood, and ongoing observations will continue to

reﬁne our understanding of the system.

One challenge is the irregularity of outbursts,

which makes long-term predictions diﬃcult. Despite

this, advancements in technology, particularly with the

next generation of space telescopes, hold great promise

for unraveling the mysteries of cataclysmic variable

systems like T CrB.

2.9 Conclusion

T Corona Borealis is an exceptional example

of a cataclysmic variable binary star system, and

its study continues to provide valuable insights into

stellar evolution, binary interactions, and the complex

dynamics of mass transfer. Through the combination of

theoretical models and observational data, astronomers

can continue to unlock the mysteries of such systems,

enhancing our understanding of the life cycle of stars

and the nature of the universe itself.

In the next article, we will explore the anticipated

nova explosion of T Coronae Borealis (T CrB), which

scientists believe could become visible around 2025.

2.9 Conclusion

However, the exact timing remains uncertain, and

researchers are eagerly monitoring this event.

References:

Corona Borealis

T Coronae Borealis

Rare T Coronae Borealis nova explosion this

September: What it is and how to see it

NASA, Global Astronomers Await Rare Nova

Explosion

T Coronae Borealis Nova Explosion 2024: Has

T CrB Nova Happened Yet?

Corona Borealis (The Northern Crown)

Constellation

Corona Borealis Constellation

Corona Borealis

A Blaze star or ‘Nova’ in Corona Borealis

A Guide to the Corona Borealis Constellation

and Its Stars

Corona Borealis

About the Author

Sindhu G is a research scholar in Physics

doing research in Astronomy & Astrophysics. Her

research mainly focuses on classiﬁcation of variable

stars using diﬀerent machine learning algorithms. She

is also doing the period prediction of diﬀerent types

of variable stars, especially eclipsing binaries and on

the study of optical counterparts of X-ray binaries.

Part III

Biosciences

Synthetic Biology: A Revolutionary Scientiﬁc

Frontier

by Geetha Paul

airis4D, Vol.3, No.2, 2025

www.airis4d.com

1.1 Introduction

Synthetic biology represents a transformative

scientiﬁc discipline that fundamentally challenges our

traditional understanding of biological systems. Unlike

conventional biological research, this innovative ﬁeld

applies sophisticated engineering principles to living

organisms, treating them as complex, programmable

systems that can be systematically designed, modiﬁed,

and reconstructed. Synthetic biology seeks to transcend

the limitations of existing biological frameworks

by creating entirely new biological components and

systems that do not naturally occur in the world.

Researchers in this ﬁeld view genetic material not

merely as a biological blueprint but as a programmable

language that can be manipulated with precision and

creativity. This approach represents a profound shift

from traditional genetic engineering, which typically

focuses on making incremental modiﬁcations to existing

organisms. The fundamental philosophy of synthetic

biology is rooted in the belief that biological systems can

be understood, standardised, and engineered, much like

mechanical or electronic systems. Scientists approach

living organisms as intricate machines composed of

interchangeable parts – or biobricks – that can be

disassembled, analysed, and reconstructed to achieve

speciﬁc, predictable functionalities. This perspective

allows researchers to design biological systems with

unprecedented levels of control and intentionality. Its

holistic and interdisciplinary approach distinguishes

synthetic biology from previous biological research.

Researchers can develop innovative solutions that

address complex challenges across multiple domains

by integrating knowledge from biology, engineering,

computer science, and computational modelling. The

ﬁeld draws inspiration from engineering principles

such as modularity, standardisation, and abstraction,

applying these concepts to biological systems in ways

that were previously unimaginable. The technological

foundations of synthetic biology are built upon advanced

tools and methodologies. Cutting-edge technologies

like CRISPR gene editing, high-throughput DNA

sequencing, and sophisticated computational modelling

enable scientists to manipulate genetic material with

unprecedented precision. Machine learning algorithms

and artiﬁcial intelligence further enhance researchers’

ability to predict and design complex biological

interactions. Unlike traditional genetic modiﬁcation,

which typically involves minor changes to existing

organisms, synthetic biology aims to create entirely

new biological functions and organisms. This approach

opens up remarkable possibilities for addressing global

challenges in healthcare, environmental sustainability,

industrial manufacturing, and beyond.

Researchers can now design microorganisms

capable of producing complex chemicals, detecting

and treating diseases, cleaning ecological pollutants,

and developing sustainable technologies. The

potential applications of synthetic biology are vast

and transformative. In medicine, scientists develop

personalised cellular therapies, create intelligent

diagnostic tools, and design targeted treatments that can

1.2 Embedded Hybridisation

adapt to individual genetic proﬁles. Environmental

researchers are engineering organisms that can

consume plastic waste, capture carbon dioxide,

and produce renewable energy sources. Industrial

biotechnologists are developing more eﬃcient and

sustainable manufacturing processes that reduce

environmental impact. However, this powerful ﬁeld

also raises important ethical considerations. The ability

to create and modify life forms necessitates careful

reﬂection on potential risks, ecological implications,

and societal consequences.

Robust regulatory frameworks and ongoing

ethical discussions are crucial to ensuring responsible

development and application of these groundbreaking

technologies. As our understanding of genetic systems

continues to deepen and technological capabilities

expand, synthetic biology promises to revolutionise

our approach to solving complex global challenges.

By treating biology as an engineering discipline,

researchers are not just modifying life but fundamentally

reimagining its potential. This ﬁeld represents

a remarkable convergence of scientiﬁc disciplines,

oﬀering unprecedented opportunities to address some

of humanity’s most pressing environmental, medical,

and technological challenges.

1.2 Embedded Hybridisation

The embedded hybridisation mode entails

the formation of hybrid cells through physical

encapsulation, which can involve either the

encapsulation of living cells within synthetic ones

or the reverse. This process is analogous to the

evolution of eukaryotic organelles, where independent

organisms were incorporated into a host, resulting

in eukaryotic cells through a symbiotic relationship.

The emerging organelles operate in a unique

physicochemical environment, enabling them to

specialise in speciﬁc functions. In artiﬁcial cell

contexts, embedded organelle-like compartments have

been utilised for various applications, such as spatially

segregated transcription and translation, stimuli-

responsive enzymatic reactions in vesicle-based cells,

and the engineering of synthetic signalling cascades

Image Courtesy: https://onlinelibrary.wiley.com/doi/full/10.1002/anie.202006941

Figure 1: Cellular bionics. Hybrid cellular bionic

systems can be constructed by fusing living and

non-living modules together. Living modules can

be cells (prokaryotic or eukaryotic) or organelles

(e.g. mitochondria and chloroplasts). Non-living

modules can consist of artiﬁcial cell-like compartments

composed of biological and synthetic molecular

components (e.g. enzymes, membrane proteins, DNA,

and nanoparticles). The construction of artiﬁcial cells

from inanimate molecular building blocks. They are

being used as simpliﬁed cell models to decipher the

rules of life. Artiﬁcial cells have the potential to be

designed as micromachines deployed in a host of clinical

and industrial applications.

1.3 The Fundamental Approach

Image Courtesy: https://onlinelibrary.wiley.com/doi/full/10.1002/anie.202006941

Figure 2: Embedded hybridisation. A–C) Examples

of living cells encapsulated in synthetic ones. A)

Engineered eukaryotic cells encapsulated in an artiﬁcial

cell containing a synthetic metabolism. Coupling

the living and synthetic cells in this way resulted

in an enzymatic cascade leading to the production

of a ﬂuorescent molecule (resoruﬁn) within the

hybrid bioreactor.B) Microscopy images showing

chloroplasts (red) encapsulated in coacervates (green).

Chloroplasts retained their light-induced electron

transport capabilities, as demonstrated by the Hill

reagent (DPIP) reduction depicted in the schematic. C)

Schematic of a chromatophore organelle extracted from

a photosynthesising organism and inserted in a synthetic

cell. Light irradiation led to the production of ATP,

which powered the translation apparatus to produce

mRNA.An example of a synthetic cell encapsulated in a

living one. The synthetic cells performed an organelle-

like function by degrading H2O2, thus shielding the

cell from the detrimental eﬀects of this molecule.

between compartments. Recently, a trend has been

growing to broaden this concept towards developing

hybrid living or synthetic systems.

1.3 The Fundamental Approach

The core philosophy of synthetic biology is to treat

biological systems like complex machines that can be

dismantled, understood, and reconstructed. Scientists in

this ﬁeld view genetic code as a programmable language,

where DNA sequences are analogous to computer code.

By breaking down biological systems into standardised,

interchangeable parts called biobricks, researchers can

design and assemble novel genetic circuits with speciﬁc

functions.

Synthetic biology seeks to design and build new

biology that does useful things.

The ﬁeld applies engineering principles of

modularity, standardisation, and abstraction

to promote rational design for industrial

applications.

DNA encodes bioparts, bioparts are combined

to make biological devices, and devices are built

into biological systems.

Synthetic biology is working towards a ‘plug-and-

play’ concept, in which oﬀ-the-shelf bioparts can

be loaded and run in ﬁnely characterised chasses.

Synthetic biology has applications in healthcare,

industrial processes, and for improving the

environment.

Creating a new life comes with a huge

responsibility. Ethical, safety, and security

implications must be considered and continually

re-assessed.

1.4 Technological and Healthcare

Innovations

Advanced technologies form the backbone of

synthetic biology. Cutting-edge tools like CRISPR

gene editing, high-throughput DNA sequencing,

and sophisticated computational modelling enable

researchers to manipulate genetic material with

unprecedented precision. Machine learning algorithms

help predict how genetic modiﬁcations might behave,

allowing scientists to design more complex and reliable

biological systems.

The potential applications of synthetic biology

are remarkably diverse and transformative. In

healthcare, researchers are developing personalised

medical treatments, creating more eﬀective drug

delivery systems, and designing diagnostic tools that can

detect diseases at extremely early stages. Environmental

scientists are engineering microorganisms capable of

consuming plastic waste, capturing carbon dioxide, or

producing sustainable biofuels.

In the medical ﬁeld, synthetic biology oﬀers

1.5 GES (General Environmental Solutions)

groundbreaking possibilities. Scientists are developing

smart cellular therapies to identify and destroy cancer

cells, create personalised medications tailored to

individual genetic proﬁles, and design biological

sensors to detect and respond to speciﬁc medical

conditions. These innovations could revolutionise

disease treatment and prevention.

Beyond healthcare, synthetic biology is

transforming industrial processes. Researchers

are creating microorganisms that can produce

complex chemicals, pharmaceuticals, and materials

more eﬃciently and sustainably than traditional

manufacturing methods. Environmental applications

include developing organisms to clean up oil spills,

convert waste into valuable resources, and create more

sustainable agricultural practices. Integrating industrial

and environmental solutions is vital for businesses

aiming to maintain competitiveness while adhering to

regulatory requirements. By focusing on sustainability

and eﬃcient resource management, companies can

comply with laws and enhance their reputation and

operational eﬀectiveness in a rapidly evolving market

landscape.

1.5 GES (General Environmental

Solutions)

GES provides environmental consulting,

engineering, compliance, and technical ﬁeld services

tailored for manufacturing and legacy industrial

contamination. They collaborate with businesses

across sectors like aerospace, chemical, pharmaceutical,

and technology to develop customised environmental

permitting, compliance, risk management, and liability

management approaches. Their integrated teams

leverage deep technical expertise to implement practical

solutions that align with business objectives.

1.6 Environmental and Industrial

Solutions (EIS)

Established in 2011, EIS specialises in halon

recovery and reclamation throughout the Middle East.

They operate the only facility in Saudi Arabia capable

of processing halons to military speciﬁcations. EIS

focuses on recycling halons from ﬁreﬁghting systems to

meet current standards while providing comprehensive

maintenance services for ﬁre protection systems.

Certain companies oﬀer a wide range of industrial

wastewater treatment equipment and chemicals. They

provide customised wastewater treatment systems,

retroﬁts, upgrades, and solidiﬁcation equipment

designed to enhance production eﬃciency while

reducing costs.

Founded in 2017, Greenlab caters to testing and

certiﬁcation needs within the construction industry.

They focus on providing superior laboratory equipment

and consultancy services while adhering to quality

standards and customer satisfaction.

Comtech oﬀers extensive consulting services in

environmental compliance alongside operations and

process management support. Their expertise helps

businesses navigate complex regulatory landscapes

while optimising their operational processes.

1.7 Ethical Considerations

As with any powerful technology, synthetic

biology raises important ethical questions. The

ability to create and modify life forms necessitates

careful consideration of potential risks, environmental

impacts, and societal implications. Robust regulatory

frameworks and ongoing ethical discussions are crucial

to ensuring responsible development and application of

these technologies.

1.8 Future Potential

The future of synthetic biology is auspicious. As

our understanding of genetic systems deepens and

technological capabilities expand, we can anticipate

increasingly sophisticated biological solutions. From

creating more resilient crops to developing novel

medical treatments, synthetic biology has the potential

to address some of humanity’s most pressing challenges.

1.9 Conclusion

Synthetic biology represents a paradigm shift

in understanding and interacting with living systems.

By treating biology as an engineering discipline,

researchers are opening up unprecedented possibilities

for solving complex global challenges. As the ﬁeld

continues to evolve, it promises to reshape our approach

to medicine, environmental conservation, industrial

production, and our fundamental understanding of life

itself.

NB: Part 2 in Next Volume .

Part 2 -Synthetic Biology-Recent advancements

in synthetic biology - transforming the ﬁeld and

paving the way for innovative applications across

various sectors. Here are some of the latest

developments:

CRISPR-Cas9 Enhancements, DNA Synthesis

Innovations, Synthetic Cells and Organs,

Synthetic Microbes,Machine Learning and

Data Science Integration, Development of Synthetic

Vaccines,Sustainable Materials Production,

Bioremediation Techniques.

References

Benner, S., & Sismour, A. (2005). Synthetic

biology. Nature Reviews Genetics, 6(6), 533–543.

https://doi.org/10.1038/nrg1637

Garner, K. L. (2021). Principles of synthetic

biology. Essays in Biochemistry, 65(5), 791-811.

https://doi.org/10.1042/EBC20200059

Wisner S. (2021) Synthetic biology investment

reached a new record of nearly $8 billion in 2020 -

what does this mean for 2021? 2020 synbiobeta market

report

https://synbiobeta.com/synthetic-biology-

investment-set-a-nearly-8-billion-record-in-2020-

what-does-this-mean-for-2021/Accessed 24/06/2021

[Google Scholar]

https://onlinelibrary.wiley.com/doi/full/10.1002/anie.202006941

About the Author

Geetha Paul is one of the directors of

airis4D. She leads the Biosciences Division. Her

research interests extends from Cell & Molecular

Biology to Environmental Sciences, Odonatology, and

Aquatic Biology.

Part IV

General

Domain Generation Algorithm (DGA)

Detection Techniques

by Jinsu Ann Mathew

airis4D, Vol.3, No.2, 2025

www.airis4d.com

Domain Generation Algorithms (DGAs) generate

domains with unique characteristics to evade traditional

detection mechanisms. In the previous article,

we explored the evolution of DGAs, tracing their

progression from simple, pattern-based generators to

more complex, AI-driven systems that make detection

even more challenging. As DGAs have become more

advanced, the techniques to detect them have also

evolved, adapting to the increasingly sophisticated

nature of these threats.

This article takes a deeper dive into the modern

techniques employed to identify and mitigate DGA-

related threats. We focus on advanced detection

methods that address the unique challenges posed by

AI-enhanced DGAs, which can generate domain names

with greater complexity and variability. Some of the

cutting-edge approaches discussed include statistical

analysis, machine learning, behavioral monitoring, and

n-gram analysis, all of which are used to detect the

telltale signs of DGA activity in real-time. These

techniques not only improve detection accuracy but

also enhance our ability to combat the ever-evolving

tactics used by cybercriminals.

1.1 Statistical Analysis

Statistical Analysis for DGA Detection focuses

on leveraging mathematical and statistical properties

of domain names to distinguish between benign and

DGA-generated domains. This approach is based on

the observation that DGA domains often have distinct

patterns that diﬀer from the typical structure of regular

domain names. The goal of statistical analysis is to

identify these diﬀerences using quantitative measures

to ﬂag potential threats.

Character Distribution

DGA domains often have a highly irregular

character distribution compared to regular domains. For

instance, in human-registered domains, the character

distribution tends to follow certain natural language

patterns, such as vowels and consonants occurring with

typical frequencies. DGA domains, on the other hand,

often feature highly random or uniform distributions

of characters, with a heavy reliance on speciﬁc letters

or symbols generated by the algorithm. Statistical

tools can analyze the frequency of characters within a

domain name, comparing it to the expected distribution

in legitimate domains. A signiﬁcant deviation can

indicate that the domain is likely generated by a DGA.

Domain Length Analysis

The length of domain names can also provide

insight into whether they are DGA-generated. Regular

domains tend to have varying lengths, but they often

fall within certain ranges (usually between 3 to 63

characters). In contrast, DGA domains may exhibit

certain length patterns—either shorter, more uniform,

or longer than typical domain names. By analyzing the

domain length distribution, statistical techniques can

ﬂag domains that fall outside of expected ranges.

1.2 Machine Learning (ML) Approaches:

Entropy Measurement

Entropy is a measure of randomness or

unpredictability in a set of data. In the context of DGA

detection, entropy is used to assess the randomness

of a domain name’s character sequence. Human-

registered domain names often have recognizable

patterns or familiar words, which results in lower

entropy. DGA domains, generated randomly or using

speciﬁc algorithms, tend to have high entropy, meaning

they are more random and unpredictable. High entropy

values are often associated with DGA domains, making

this a useful statistical measure for detection.

1.2 Machine Learning (ML)

Approaches:

Machine Learning (ML) approaches for Domain

Generation Algorithm (DGA) detection have gained

prominence due to their ability to automatically learn

from data and adapt to evolving patterns, especially

when dealing with the more sophisticated AI-driven

DGAs. These techniques involve using labeled datasets

of both benign and DGA-generated domains to train

models that can then predict whether an unseen domain

is benign or malicious.

Supervised Learning:

Supervised learning relies on labeled datasets

containing both benign and DGA domains. Features

such as domain length, character frequency, entropy,

n-grams, and domain suﬃxes are extracted from domain

names to train models like decision trees, random

forests, support vector machines (SVM), or neural

networks. Once trained, these models classify new

domains with high accuracy, making them ideal for real-

time detection systems. However, their eﬀectiveness

depends on having a diverse and comprehensive labeled

dataset.

Unsupervised Learning:

When labeled data is limited, unsupervised

learning becomes useful. Techniques like clustering

(e.g., K-means or DBSCAN) group domains with

similar characteristics, while anomaly detection

algorithms (e.g., Isolation Forest or One-Class SVM)

identify domains that deviate signiﬁcantly from normal

patterns. These methods are particularly eﬀective for

detecting previously unknown DGAs.

Deep Learning:

Deep learning models, such as Recurrent Neural

Networks (RNNs) and Long Short-Term Memory

(LSTM) networks, process domain names as sequences

of characters, capturing intricate patterns in their

structure. These models can automatically learn feature

representations, eliminating the need for manual feature

extraction. Convolutional Neural Networks (CNNs)

have also been applied, treating domains as sequences

where local patterns can be identiﬁed. Deep learning

excels in handling complex and dynamic DGAs but

requires large datasets and signiﬁcant computational

resources.

Hybrid Approaches:

Hybrid approaches in DGA detection combine

the strengths of multiple machine learning techniques,

leveraging both supervised and unsupervised methods

to enhance accuracy, adaptability, and robustness. This

integration is particularly valuable in scenarios where

labeled data is incomplete or when new, unseen DGA

patterns emerge.

1.3 N-gram Analysis

N-gram Analysis is a powerful technique for

detecting domains generated by Domain Generation

Algorithms (DGAs). This method examines the

sequences of characters, or ”N-grams,” within

domain names to identify patterns that distinguish

human-generated domains from machine-generated

ones. Human-created domains, such as ”google.com”

or ”example.org,” follow predictable linguistic and

semantic structures. In contrast, DGA-generated

domains like ”xkzji.com” often appear random or

1.4 DNS Query Patterns

unnatural, making them detectable through their distinct

N-gram patterns.

The process begins by extracting N-grams from

domain names, where an N-gram represents a sequence

of ’n’ consecutive characters. For example, in the

domain ”domain.com,” the bigrams (2-grams) include

”do,” ”om,” and ”ai.” The frequency of these N-

grams is then compared to a baseline constructed from

large datasets of legitimate domains. By identifying

unusual or statistically improbable N-gram distributions,

suspicious domains can be ﬂagged as potentially

malicious.

To enhance accuracy, the extracted N-gram

features are often fed into machine learning models

trained to classify domains as benign or DGA-generated.

Advanced techniques, such as weighting rare N-

grams more heavily or using sliding windows to

capture overlapping patterns, further reﬁne the analysis.

Additionally, language models like Markov Chains can

predict the likelihood of an N-gram sequence, with

lower probabilities signaling potential DGA activity.

N-gram analysis is scalable, eﬃcient, and

adaptable to diﬀerent languages and scripts, making

it an essential tool for real-time detection. However,

modern DGAs, especially those powered by AI, can

mimic natural language patterns, posing challenges to

detection. Despite these challenges, N-gram analysis

remains a cornerstone of DGA detection and is often

integrated with other methods, such as DNS traﬃc

monitoring, to enhance overall eﬃcacy.

1.4 DNS Query Patterns

DNS query pattern analysis is a powerful method

used to detect domains generated by Domain Generation

Algorithms (DGAs). DGAs often rely on generating a

vast number of domains to communicate with Command

and Control (C&C) servers, creating distinct patterns in

DNS queries that can be analyzed to identify malicious

activity.

Identifying Anomalous Query Frequencies

One of the key characteristics of DGA activity is

the generation of a large number of domain names, only

a small subset of which are registered and actively

used by attackers. This behavior leads to a high

volume of DNS queries, most of which result in

failed resolutions, known as NXDOMAIN responses

(non-existent domain). By analyzing DNS query

logs, security systems can detect patterns of unusually

frequent NXDOMAIN responses, which often indicate

DGA activity. Additionally, DGAs may generate

domains using uncommon or rarely observed Top-Level

Domains (TLDs), further standing out from legitimate

domain traﬃc. Monitoring these high query rates and

unusual TLDs allows security systems to ﬂag suspicious

behavior and isolate potential DGA-generated domains.

Monitoring Temporal Patterns

Many DGAs operate on time-based generation

algorithms, creating new domains periodically based

on a predeﬁned algorithm. This results in distinctive

temporal patterns in DNS queries. For instance, a bot

infected with a DGA-based malware may attempt to

resolve speciﬁc domains at consistent intervals, aligning

with the algorithm’s time-based logic. By analyzing

DNS query logs over time, it becomes possible to

identify these periodic behaviors. Domains queried in

tightly packed bursts or those that follow a predictable

daily or hourly pattern can be strong indicators of DGA

activity, as legitimate domain queries rarely exhibit

such consistent temporal regularity.

Identifying Newly Observed and Random

Domains

DGA-generated domains often diﬀer signiﬁcantly

from legitimate domains in their structure and usage

history. Most legitimate domains have a track record

of DNS queries and resolutions over time. In contrast,

DGA domains are newly generated and queried for

the ﬁrst time, making them stand out in DNS traﬃc

logs. Additionally, DGA-generated domains often

appear random, with unusual character distributions

1.5 Conclusion

or nonsensical arrangements. By ﬂagging domains

that are newly observed or lack meaningful resolution

history, security systems can isolate potential DGA

domains for further investigation.

1.5 Conclusion

In this article, we explored the diverse and

advanced techniques used to detect domains generated

by Domain Generation Algorithms (DGAs). As DGAs

have evolved from simple, rule-based approaches to

complex, AI-driven systems, detection strategies have

also advanced to counter these threats eﬀectively.

Statistical analysis, machine learning, n-gram analysis,

and DNS query pattern analysis each play a critical role

in identifying and mitigating DGA-based activities.

Statistical methods help uncover anomalies in

domain characteristics and query frequencies, oﬀering

a foundational layer of detection. Machine learning

techniques provide robust, adaptive solutions by training

models to identify intricate patterns indicative of

DGA behavior. N-gram analysis, focusing on the

structural patterns within domain names, excels at

identifying linguistic irregularities and randomness in

DGA-generated domains. DNS query pattern analysis,

with its focus on anomalous query rates, temporal

behaviors, and network-wide correlations, oﬀers real-

time capabilities for detecting malicious activities at

scale.

Together, these approaches form a comprehensive

toolkit for combating the sophisticated and ever-

evolving strategies employed by cybercriminals. The

integration of multiple detection techniques ensures

a layered and adaptive defense system, capable of

addressing the unique challenges posed by AI-enhanced

DGAs. Moving forward, continued advancements

in detection strategies, coupled with interdisciplinary

research, will be essential in safeguarding networks and

systems against this persistent cybersecurity threat.

References

Explained: Domain Generating Algorithm

Domain generation algorithm

Detecting DGA Domains: Machine Learning

Approach

DGA Detection with data analytics

Detecting DGA domains with recurrent neural

networks and side information

Real-Time Detection of Dictionary DGA Network

Traﬃc Using Deep Learning

About the Author

Jinsu Ann Mathew is a research scholar

in Natural Language Processing and Chemical

Informatics. Her interests include applying basic

scientiﬁc research on computational linguistics,

practical applications of human language technology,

and interdisciplinary work in computational physics.

About airis4D

Artiﬁcial Intelligence Research and Intelligent Systems (airis4D) is an AI and Bio-sciences Research Centre.

The Centre aims to create new knowledge in the ﬁeld of Space Science, Astronomy, Robotics, Agri Science,

Industry, and Biodiversity to bring Progress and Plenitude to the People and the Planet.

Vision

Humanity is in the 4th Industrial Revolution era, which operates on a cyber-physical production system. Cutting-

edge research and development in science and technology to create new knowledge and skills become the key to

the new world economy. Most of the resources for this goal can be harnessed by integrating biological systems

with intelligent computing systems oﬀered by AI. The future survival of humans, animals, and the ecosystem

depends on how eﬃciently the realities and resources are responsibly used for abundance and wellness. Artiﬁcial

intelligence Research and Intelligent Systems pursue this vision and look for the best actions that ensure an

abundant environment and ecosystem for the planet and the people.

Mission Statement

The 4D in airis4D represents the mission to Dream, Design, Develop, and Deploy Knowledge with the ﬁre of

commitment and dedication towards humanity and the ecosystem.

Dream

To promote the unlimited human potential to dream the impossible.

Design

To nurture the human capacity to articulate a dream and logically realise it.

Develop

To assist the talents to materialise a design into a product, a service, a knowledge that beneﬁts the community

and the planet.

Deploy

To realise and educate humanity that a knowledge that is not deployed makes no diﬀerence by its absence.

Campus

Situated in a lush green village campus in Thelliyoor, Kerala, India, airis4D was established under the auspicious

of SEED Foundation (Susthiratha, Environment, Education Development Foundation) a not-for-proﬁt company

for promoting Education, Research. Engineering, Biology, Development, etc.

The whole campus is powered by Solar power and has a rain harvesting facility to provide suﬃcient water supply

for up to three months of drought. The computing facility in the campus is accessible from anywhere through a

dedicated optical ﬁbre internet connectivity 24×7.

There is a freshwater stream that originates from the nearby hills and ﬂows through the middle of the campus.

The campus is a noted habitat for the biodiversity of tropical Fauna and Flora. airis4D carry out periodic and

systematic water quality and species diversity surveys in the region to ensure its richness. It is our pride that the

site has consistently been environment-friendly and rich in biodiversity. airis4D is also growing fruit plants that

can feed birds and provide water bodies to survive the drought.