This paper analyzes a relay communication system with a
single source, N relays, and a direct source–destination
link under Rayleigh fading conditions. Amplify-and-Forward
(AF) relaying and Selection Combining (SC) at the receiver
are employed, while the system outage probability (Pout) is
evaluated via statistical analysis of the instantaneous
signal-to-noise ratio (SNR). The results indicate that the
receiver sensitivity threshold (γth) has a dominant effect
on system performance, whereas the relay transmit power has
a smaller impact. Additionally, the study explores the use
of locally deployable Generative Artificial Intelligence
(GenAI) for agentic, automated network design and
management, showing reduced execution time and effort while
maintaining accuracy, highlighting its potential for
efficient 5G/6G network optimization. On the other side,
adoption of agentic memory and Model Context Protocol (MCP)
with locally deployable model results with outcomes
comparable to commercial, externally managed GenAI services.

This paper investigates a wireless communication system
employing an L-branch selection combiner (SC), where the
input signals are subject to Weibull fading, and co-channel
interference (CCI) is also considered, modeled using the
Weibull distribution. To mitigate the effects of fading and
interference, SC is applied as a simple yet effective
diversity technique. In the first part of the paper, a
closed-form expression for the level crossing rate (LCR) of
the signal-to-interference ratio (SIR) is derived,
providing analytical insight into the system dynamics.
Numerical results demonstrate how the fading and
interference parameters influence the LCR. In the second
part, we show how the obtained values can be further
utilized in an LLM-driven approach for network management,
enabling evaluation of functional safety aspects and
performance optimization of the network configuration.

In this paper, we examine the behavior of the optimal
factorization framework for multistage decimation
introduced in our earlier work. The objective is not to
modify the optimization procedure, but to clarify how the
optimal realizations are distributed across the
decimation-ratio range and how sensitive the achieved gains
are to the filter specifications. Using a computation rate
model based on multiplications per second, we compare
globally optimal realizations based on the transposed
modified Farrow structure (TMFS) followed by FIR stages
with the corresponding FIR-only cascades. The results show
a clear separation between prime and composite ratios and
confirm that the optimal structures remain shallow across
the considered range. A sensitivity analysis further
indicates that the passband edge has the strongest
influence on the achievable gain. These results complement
the earlier algorithmic contribution and provide practical
guidance on when the generalized TMFS–FIR architecture is
the most beneficial.

This paper presents the design and implementation of an
intra-node control communication framework for modular
Optical Transport Network (OTN) systems. The proposed
architecture coordinates a Central Management Unit (CMU)
and distributed line card applications within an industrial
OTN platform. Communication is realized through a
lightweight message-oriented layer based on ZeroMQ,
combining Request–Reply transactions for deterministic
configuration handling and Publisher–Subscriber signaling
for asynchronous status exchange. JSON is used for
structured and extensible message representation, while
systemd provides supervised service lifecycle management
and fault isolation. The framework has been implemented and
validated on an OTN node platform developed by Iritel.

With rapid expansion of IoT applications, ensuring adequate
security for IoT systems has become increasingly
challenging. Blockchain technology has proven to be very
effective at protecting data integrity and has been
researched as a promising addition to network security,
across various industries. It is characterized by a
transparent and immutable nature, while relying on
decentralized and distributed networks, as well as
cryptographic mechanisms to achieve a high level of
security. This paper proposes a network architecture that
utilizes Blockchain technology, along with a voting system
and a consensus model based on random role assignment, to
establish a secure environment for a typical IoT topology.
By leveraging decentralized storage of security records,
the proposed approach eliminates single points of failure,
while enabling faster and more seamless device
authentication.

In modern optical transport networks, new requirements
arise that need to be met with appropriate equipment. The
paper presents new solutions in the architecture of Optical
Transport Network units with a focus on the central control
system and communication with the Ethernet chip. In order
to implement the aforementioned communication, it is
necessary to perform a successful hardware connection,
adapt the configurable logic and ensure the entire
communication chain between the user space of the operating
system and the hardware. In addition to demonstrating the
verification of the functionality of the implemented
interface, technical details about the development platform
and the applied technologies are also presented.

This paper presents the design of an MDIO controller
implemented on Microchip Polarfire SoC platform. The
proposed controller is integrated in Icicle open source
project that contains RISC-V processor with different types
of peripherals. The proposed peripheral is APB slave to the
processosor and uses MDIO 45 clause interface to
communicate with CFP2 optical transciever. Operating system
that is used in Linux Ubuntu and there is a driver that is
written in Iritel that drives the peripheral. The
implementation and the results of the custom solution will
be presented in the following text with the comparison to
the Microchip core component from the catalog (MDIO_APB),
also integrated in Icicle project.

Kvalitet vazduha u zatvorenom prostoru predstavlja značajan
faktor rizika po zdravlje, jer ljudi provode najveći deo
svog vremena u zatvorenim prostorima poput stanova, škola,
radnih mesta i kancelarija. U ovom radu prikazan je
praktičan pristup za kontinuirano praćenje parametara
kvaliteta vazduha u zatvorenom, koristeći prenosivi uređaj
sa niskobudžetnim senzorima i IoT tehnologijom za
prikupljanje i analizu podataka. Merenja obuhvataju
koncentracije suspendovanih čestica (PM1, PM2.5, PM10),
isparljiva organska jedinjenja (VOC), temperaturu, pritisak
i relativnu vlažnost vazduha. Podaci su prikupljani u
različitim tipičnim okruženjima (stambeni prostori, sala za
trening, kupatilo) kako bi se uočile promene usled
uobičajenih aktivnosti. Rezultati pokazuju da izlaganje
duvanskom dimu, upotreba aerosola i nedovoljna ventilacija
dovode do osetnog porasta PM i VOC indikatora, dok
provetravanje uglavnom smanjuje VOC i stabilizuje PM
vrednosti. Ostvareni rezultati postavljaju okvire za buduća
istraživanja sa ciljem da se pravovremeno prepozna
nepovoljan kvalitet vazduha u zatvorenom prostoru, razviju
smernice za frekventnost i trajanje provetravanja i
edukaciju korisnika o uticaju svakodnevnih aktivnosti na
kvalitet vazduha.

U radu je predstavljeno modelovanje i simulacija MEMS
(Micro-Electro-Mechanical Systems) pito-statičkog sistema
za estimaciju vazdušnih parametara u uslovima degradiranih
merenja. Analiziran je uticaj šuma, bias drifta, dinamičkih
efekata i zaleđivanja Pito cevi. Za estimaciju stanja
primenjen je prošireni Kalmanov filter (EKF, Extended
Kalman filter), uz adaptivni pristup zasnovan naNormalized
Innovation Squared (NIS) kriterijumu za detekciju anomalija
i za prilagođavanje matrice šuma merenja. Rezultati
simulacije pokazuju da primena EKF daje dobre rezultate u
nominalnim uslovima, dok primenu adaptivnog EKF odlikuje
značajno smanjivanje greške procene u prisustvu poremećaja,
posebno za merenje brzine. Predloženi pristup unapređuje
pouzdanost sistema za obradu vazdušnih podataka u realnim
uslovima eksploatacije.