Adrian Peculea

Dr. Adrian Peculea

Technical University of Cluj-Napoca, Computer Science Department
orcid

Research Focus & Metrics

Academic Career Length

20Years (2007-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
24 August 2020 in Sensors
Verified
Journal Article
Open access
PDF file
SBSorin Buzura
Sorin Buzura
BIBogdan Iancu
Bogdan Iancu
Vasile Teodor DADARLATVasile Teodor DADARLAT
Vasile Teodor DADARLAT
APAdrian Peculea
Adrian Peculea
ECEmil Cebuc
Emil Cebuc
Software-defined wireless sensor networking (SDWSN) is an emerging networking architecture which is envisioned to become the main enabler for the internet of things (IoT). In this architecture, the sensors plane is managed by a control plane. With this separation, the network management is facilitated, and performance is improved in dynamic environments. One of the main issues a sensor environment is facing is the limited lifetime of network devices influenced by high levels of energy consumption. The current work proposes a system design which aims to improve the energy efficiency in an SDWSN by combining the concepts of content awareness and adaptive data broadcast. The purpose is to increase the sensors’ lifespan by reducing the number of generated data packets in the resource-constrained sensors plane of the network. The system has a distributed management approach, with content awareness being implemented at the individual programmable sensor level and the adaptive data broadcast being performed in the control plane. Several simulations were run on historical weather and the results show a significant decrease in network traffic. Compared to similar work in this area which focuses on improving energy efficiency with complex algorithms for routing, clustering, or caching, the current proposal employs simple computing procedures on each network device with a high impact on the overall network performance.

26 February 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
RKRudolf Kovacs
Rudolf Kovacs
SBSorin Buzura
Sorin Buzura
BIBogdan Iancu
Bogdan Iancu
Vasile Teodor DADARLATVasile Teodor DADARLAT
Vasile Teodor DADARLAT
APAdrian Peculea
Adrian Peculea
ECEmil Cebuc
Emil Cebuc
The network function virtualization (NFV) feature lies at the core of modern networking, and it allows on-demand real-time integration of new network functions, which is a great benefit for large-scale infrastructure networks. In contrast to the functional benefits, NFV introduces software complexity and computational overhead through additional abstraction layers. The current article addresses the function validation problem in large-scale infrastructure networks of Internet Service Providers (ISPs) and proposes the utilization of blockchain as a validation technology, as opposed to implementing a custom validation solution. The current work showcases a practical architecture implementation to address the service validation in service provider large-scale networks. The POX-based solution to control software-defined networks (SDN) for NFV is extended to offer additional blockchain capabilities. Thus, a blockchain node is integrated and executed in the POX SDN controller. Transaction experiments are performed between two endpoints located in remote locations on the Internet, and the detailed results are presented to validate the utilization of the blockchain technology used on SDNs’ control plane.

2 January 2023 in Sensors
Verified
Journal Article
Open access
PDF file
SBSorin Buzura
Sorin Buzura
APAdrian Peculea
Adrian Peculea
BIBogdan Iancu
Bogdan Iancu
ECEmil Cebuc
Emil Cebuc
Vasile Teodor DADARLATVasile Teodor DADARLAT
Vasile Teodor DADARLAT
RKRudolf Kovacs
Rudolf Kovacs
In recent years, the software-defined networking (SDN) paradigm has been deployed in various types of networks, including wireless sensor networks (WSN), wide area networks (WAN) and data centers. Given the wide range of SDN domain applicability and the large-scale environments where the paradigm is being deployed, creating a full real test environment is a complex and costly task. To address these problems, software-based simulations are employed to validate the proposed solutions before they are deployed in real networks. However, simulations are constrained by relying on replicating previously saved logs and datasets and do not use real time hardware data. The current article addresses this limitation by creating a novel hybrid software and hardware SDN simulation testbed where data from real hardware sensors are directly used in a Mininet emulated network. The article conceptualizes a new approach for expanding Mininet’s capabilities and provides implementation details on how to perform simulations in different contexts (network scalability, parallel computations and portability). To validate the design proposals and highlight the benefits of the proposed hybrid testbed solution, specific scenarios are provided for each design idea. Furthermore, using the proposed hybrid testbed, new datasets can be easily generated for specific scenarios and replicated in more complex research.

1 May 2020
Verified
Conference Paper
Closed access
SBSorin Buzura
Sorin Buzura
Vasile Teodor DADARLATVasile Teodor DADARLAT
Vasile Teodor DADARLAT
BIBogdan Iancu
Bogdan Iancu
APAdrian Peculea
Adrian Peculea
ECEmil Cebuc
Emil Cebuc
RKRudolf Kovacs
Rudolf Kovacs
This paper proposes a self-adaptive algorithm for improving the quality of service (QoS) rules in the distributed control plane of a software defined network (SDN). The algorithm is based on fuzzing techniques and requires that the Controllers collaborate in order to determine the best rules to use for packet forwarding. Several simulations were run with different network configurations to verify the algorithm's effectiveness by measuring the packet miss ratio on the data plane switches. The results show that a distributed and collaborative approach is generating the set of rules with the lowest miss ratio as well as it is the most efficient in decreasing the miss ratio with each algorithm iteration, compared to other approaches which are non distributed or non collaborative.

15 July 2020 in Journal of Educational Sciences
Verified
Journal Article
Open access
LPLorena Peculea
Lorena Peculea
APAdrian Peculea
Adrian Peculea

30 March 2024 in Data
Verified
Journal Article
Open access
PDF file
RBRadu Bouaru
Radu Bouaru
APAdrian Peculea
Adrian Peculea
BIBogdan Iancu
Bogdan Iancu
SBSorin Buzura
Sorin Buzura
ECEmil Cebuc
Emil Cebuc
Vasile Teodor DADARLATVasile Teodor DADARLAT
Vasile Teodor DADARLAT
In academic environments, students take exams simultaneously in campus examination classrooms. Due to recent advancements in technology, examination rooms are flooded with Bluetooth data traffic generated by personal devices (smartphones, smartwatches, etc.). The work presented in this article proposes a method for collecting Bluetooth traffic in an academic examination setting. The desired data were collected during several examination sessions using an Ubertooth One device, and then an in-depth post-processing analysis was performed on the collected dataset. The devices generating traffic were precisely located within the examination room, and areas with heightened data traffic were highlighted. Additionally, another goal of the current research was to provide a unique type of dataset to the academic community, facilitating its utilization in further research endeavors.