OuluREPO University of Oulu repository
Not a member yet
54977 research outputs found
Sort by
Beyond Reinforcement Learning for network security: A comprehensive survey and tutorial
No full textAbstract
Maintaining strong security is a complex yet vital challenge in the rapidly evolving landscape of modern digital networks. The risks and consequences of security breaches make neglecting network protection unacceptable. Fortunately, ongoing advances in computer science have equipped researchers with powerful tools to reinforce network defenses. Among these, Reinforcement Learning (RL), a branch of machine learning, has gained significant attention for its versatility and effectiveness in strengthening security mechanisms. This paper presents a comprehensive survey and tutorial on the role of RL in network security. It provides background information, a step-by-step tutorial for training RL models, and systematically categorizes research efforts based on the targeted cyber threats. Leveraging recent advances and real-world applications, this survey elucidates how RL enables the development of adaptive and intelligent systems that autonomously learn and respond to evolving threats. Through in-depth analysis, we provide a comprehensive view of the current landscape and the future potential of RL in safeguarding digital assets. The main contributions of this survey are: (1) a systematic and up-to-date review of RL approaches for network security; (2) a unified taxonomy for classifying RL-based solutions; (3) a comparison of the latest advances from 2019 to 2024 across mainstream and emerging research areas; (4) identification of open challenges and future research directions; and (5) a comparative analysis of state-of-the-art models, offering practical insights for both researchers and practitioners. Furthermore, this survey emphasizes the practical translation of RL advances into real-world deployments. By focusing on hands-on implementation guidelines and comparative analyses of deployment scenarios, it bridges the gap between academic research and operational practice. The comprehensive evaluation of RL-based models across different network environments provides actionable insights for practitioners seeking adaptive and scalable security solutions in dynamic and heterogeneous settings.Abstract
Maintaining strong security is a complex yet vital challenge in the rapidly evolving landscape of modern digital networks. The risks and consequences of security breaches make neglecting network protection unacceptable. Fortunately, ongoing advances in computer science have equipped researchers with powerful tools to reinforce network defenses. Among these, Reinforcement Learning (RL), a branch of machine learning, has gained significant attention for its versatility and effectiveness in strengthening security mechanisms. This paper presents a comprehensive survey and tutorial on the role of RL in network security. It provides background information, a step-by-step tutorial for training RL models, and systematically categorizes research efforts based on the targeted cyber threats. Leveraging recent advances and real-world applications, this survey elucidates how RL enables the development of adaptive and intelligent systems that autonomously learn and respond to evolving threats. Through in-depth analysis, we provide a comprehensive view of the current landscape and the future potential of RL in safeguarding digital assets. The main contributions of this survey are: (1) a systematic and up-to-date review of RL approaches for network security; (2) a unified taxonomy for classifying RL-based solutions; (3) a comparison of the latest advances from 2019 to 2024 across mainstream and emerging research areas; (4) identification of open challenges and future research directions; and (5) a comparative analysis of state-of-the-art models, offering practical insights for both researchers and practitioners. Furthermore, this survey emphasizes the practical translation of RL advances into real-world deployments. By focusing on hands-on implementation guidelines and comparative analyses of deployment scenarios, it bridges the gap between academic research and operational practice. The comprehensive evaluation of RL-based models across different network environments provides actionable insights for practitioners seeking adaptive and scalable security solutions in dynamic and heterogeneous settings
Preoperative MRCP Can Rule Out Choledocholithiasis in Acute Cholecystitis with a High Negative Predictive Value: Prospective Cohort Study with Intraoperative Cholangiography
No full textAbstract
Background:
Magnetic resonance cholangiopancreatography (MRCP) provides a noninvasive and fast modality for imaging the biliary tree when choledocholithiasis is suspected. Guidelines suggest that MRCP is recommended when strong or moderate signs of common bile duct (CBD) stones are present. Well-performed prospective studies are scarce regarding the sensitivity and specificity of preoperative MRCP in patients with acute cholecystitis in comparison with intraoperative cholangiography, ERCP, or choledochoscopy.
Methods:
We performed a prospective, observational population-based feasibility study in Central Finland Hospital Nova between January 2019 and December 2019. We examined the diagnostic performance of preoperative MRCP on consecutive patients with acute cholecystitis scheduled for index admission cholecystectomy. The accuracy of MRCP was verified with IOC, choledochoscopy, or ERCP. The interobserver reliability of the image quality of MRCP and the sensitivity and specificity of choledocholithiasis were observed independently by three experienced radiologists.
Results:
A total of 180 consecutive patients diagnosed with acute cholecystitis followed by index admission cholecystectomy were identified. MRCP was performed in 113/180 (62.8%) patients, and complementary perioperative imaging of the bile ducts was performed in 72/113 (63.7%) patients. The incidence of choledocholithiasis was high (29.2%). In acute cholecystitis, the sensitivity (76.2–85.7%) and specificity (84.3–92.2%) of MRCP were equally compared to the literature with unselected patient groups. The best visibility was observed in the common hepatic duct, the inferior CBD, and the central hepatic duct. The interobserver reliability was excellent for determining the size and quantity of CBD stones.
Conclusion:
In acute cholecystitis, MRCP yields high negative predictive value regarding detection of choledocholithiasis. If CBD stones were discovered, the interobserver reliability was excellent when measuring the size and number of CBD stones. The best-visualized area was the distal part of the biliary tract, which provides good preoperative workup if choledocholithiasis is present.Abstract
Background:
Magnetic resonance cholangiopancreatography (MRCP) provides a noninvasive and fast modality for imaging the biliary tree when choledocholithiasis is suspected. Guidelines suggest that MRCP is recommended when strong or moderate signs of common bile duct (CBD) stones are present. Well-performed prospective studies are scarce regarding the sensitivity and specificity of preoperative MRCP in patients with acute cholecystitis in comparison with intraoperative cholangiography, ERCP, or choledochoscopy.
Methods:
We performed a prospective, observational population-based feasibility study in Central Finland Hospital Nova between January 2019 and December 2019. We examined the diagnostic performance of preoperative MRCP on consecutive patients with acute cholecystitis scheduled for index admission cholecystectomy. The accuracy of MRCP was verified with IOC, choledochoscopy, or ERCP. The interobserver reliability of the image quality of MRCP and the sensitivity and specificity of choledocholithiasis were observed independently by three experienced radiologists.
Results:
A total of 180 consecutive patients diagnosed with acute cholecystitis followed by index admission cholecystectomy were identified. MRCP was performed in 113/180 (62.8%) patients, and complementary perioperative imaging of the bile ducts was performed in 72/113 (63.7%) patients. The incidence of choledocholithiasis was high (29.2%). In acute cholecystitis, the sensitivity (76.2–85.7%) and specificity (84.3–92.2%) of MRCP were equally compared to the literature with unselected patient groups. The best visibility was observed in the common hepatic duct, the inferior CBD, and the central hepatic duct. The interobserver reliability was excellent for determining the size and quantity of CBD stones.
Conclusion:
In acute cholecystitis, MRCP yields high negative predictive value regarding detection of choledocholithiasis. If CBD stones were discovered, the interobserver reliability was excellent when measuring the size and number of CBD stones. The best-visualized area was the distal part of the biliary tract, which provides good preoperative workup if choledocholithiasis is present
Representation and objective reality
No full textAbstract
This chapter investigates Descartes’ theory of ideas by focusing on the distinctions he draws between different functions. It argues that Descartes is committed to a dual function of ideas: unifying the mind with its object and providing psychological and epistemic access to that object. Drawing on an analysis of Descartes’ terminology—especially his use of the terms “material,” “objective,” and “formal”—this chapter reconstructs a trichotomy underlying his conception of ideas. It aims to show how each term corresponds to a distinct perspective: ideas as mental operations, as unifications with objects, and as representations subject to truth and falsity. This framework is used to reinterpret Descartes’ responses to critics such as Arnauld and Desgabets and to reassess Margaret Wilson’s influential claim that Descartes’ view collapses into incoherence. This chapter ultimately defends the coherence of Descartes’ position by distinguishing misrepresentation from misattribution and by showing how the special status of the cogito reveals a case in which representation and objective reality converge, eliminating the possibility of error.Abstract
This chapter investigates Descartes’ theory of ideas by focusing on the distinctions he draws between different functions. It argues that Descartes is committed to a dual function of ideas: unifying the mind with its object and providing psychological and epistemic access to that object. Drawing on an analysis of Descartes’ terminology—especially his use of the terms “material,” “objective,” and “formal”—this chapter reconstructs a trichotomy underlying his conception of ideas. It aims to show how each term corresponds to a distinct perspective: ideas as mental operations, as unifications with objects, and as representations subject to truth and falsity. This framework is used to reinterpret Descartes’ responses to critics such as Arnauld and Desgabets and to reassess Margaret Wilson’s influential claim that Descartes’ view collapses into incoherence. This chapter ultimately defends the coherence of Descartes’ position by distinguishing misrepresentation from misattribution and by showing how the special status of the cogito reveals a case in which representation and objective reality converge, eliminating the possibility of error
A multi-physical field coupling modeling incorporated the non-thermal effect: carbothermic reduction of ZnFe2O4 by using microwave heating
No full textAbstract
Microwave heating has recently gained significant attention and is increasingly applied in the pyrometallurgical recovery of Zn (ZnFe2O4) from electric arc furnace dust (EAFD). Under microwave irradiation, a portion of the microwave energy is stored within molecules, enhancing the carbothermic reduction (non-thermal effect). However, quantifying the influence of the non-thermal effect in experiments remains challenging due to the invisibility of microwaves. For advancing the scalability of microwave-assisted EAFD processing, the present study focuses on ZnFe2O4 and develops a multiphysics model incorporating the non-thermal effect to elucidate the enhancement mechanism of microwaves on the reduction process. The main conclusions indicate that increasing microwave input power and graphite addition effectively enhances both the heating and reaction efficiency of the ZnFe2O4-graphite mixture. Higher power input improves efficiency while mitigating thermal runaway risks more effectively than increasing graphite addition. However, the uneven distribution of microwaves leads to localized thermal effects, impacting heating uniformity and overall reaction consistency within the mixture. Furthermore, non-isothermal kinetic analysis reveals that microwave irradiation enhances the carbothermic reduction of ZnFe2O4 through both thermal and non-thermal effects. With a stoichiometric coefficient of 1.2 for graphite addition, localized thermal and non-thermal effects reduce the apparent activation energy by 34.3 % and 25.8 %, respectively.Abstract
Microwave heating has recently gained significant attention and is increasingly applied in the pyrometallurgical recovery of Zn (ZnFe2O4) from electric arc furnace dust (EAFD). Under microwave irradiation, a portion of the microwave energy is stored within molecules, enhancing the carbothermic reduction (non-thermal effect). However, quantifying the influence of the non-thermal effect in experiments remains challenging due to the invisibility of microwaves. For advancing the scalability of microwave-assisted EAFD processing, the present study focuses on ZnFe2O4 and develops a multiphysics model incorporating the non-thermal effect to elucidate the enhancement mechanism of microwaves on the reduction process. The main conclusions indicate that increasing microwave input power and graphite addition effectively enhances both the heating and reaction efficiency of the ZnFe2O4-graphite mixture. Higher power input improves efficiency while mitigating thermal runaway risks more effectively than increasing graphite addition. However, the uneven distribution of microwaves leads to localized thermal effects, impacting heating uniformity and overall reaction consistency within the mixture. Furthermore, non-isothermal kinetic analysis reveals that microwave irradiation enhances the carbothermic reduction of ZnFe2O4 through both thermal and non-thermal effects. With a stoichiometric coefficient of 1.2 for graphite addition, localized thermal and non-thermal effects reduce the apparent activation energy by 34.3 % and 25.8 %, respectively
The Finnish‑Swedish borderland as a resilient space of cross‑border relations
No full textAbstract
Swedish borderland populations become spatially socialized within their “own” county, its territory and identity narratives. Second, the European Union membership of Finland and Sweden in 1995 in entailed a major shift in the borderlands, facilitating cooperation and infrastructures across the border. It is investigated how the state-centric trajectory has been shaped and, in many ways, challenged by the multilevel governance structures of the European Union and related cross-border regionalizations as well as the tradition of Nordic cooperation. The third area of investigation concerns the manifestation of borderland resilience, contemplating what makes the Finnish-Swedish borderland a specific place of dwelling, identifying with it. It shows that although the European “long summer of migration” of 2015 and the state-led COVID-19 border restrictions turned down many institutionalized forms of cross-border mobility and cooperation, at least temporarily, cross-border relations and borderland identities also indicate resilience. Borderlands are both spaces of contact and divisions. More research is needed on different European borderlands to understand to what extend different cross-border relations and identities are resilient to border transitions and geopolitical change.Abstract
Swedish borderland populations become spatially socialized within their “own” county, its territory and identity narratives. Second, the European Union membership of Finland and Sweden in 1995 in entailed a major shift in the borderlands, facilitating cooperation and infrastructures across the border. It is investigated how the state-centric trajectory has been shaped and, in many ways, challenged by the multilevel governance structures of the European Union and related cross-border regionalizations as well as the tradition of Nordic cooperation. The third area of investigation concerns the manifestation of borderland resilience, contemplating what makes the Finnish-Swedish borderland a specific place of dwelling, identifying with it. It shows that although the European “long summer of migration” of 2015 and the state-led COVID-19 border restrictions turned down many institutionalized forms of cross-border mobility and cooperation, at least temporarily, cross-border relations and borderland identities also indicate resilience. Borderlands are both spaces of contact and divisions. More research is needed on different European borderlands to understand to what extend different cross-border relations and identities are resilient to border transitions and geopolitical change
Viscosity and structure of low basicity CaO-SiO2–10wt.% FeOx-10wt.% P2O5 slag system
No full textAbstract
Slag viscosity is a crucial factor affecting the dissolution of silica modifiers in molten basic oxygen furnace (BOF) slag, particularly during the online modification process. In this work, the viscosity of low basicity CaO-SiO2–10 wt.% Fe2O3–10 wt.% P2O5 slag was measured by the rotating cylinder method. The structure of glassy slag samples was analyzed by multiple spectroscopic techniques to offer a better understanding of the relationship between slag viscosity and structure. It is concluded that the slag viscosity increases with decreasing binary basicity (CaO wt.%/SiO2 wt.%) from 1.46 to 1.16, and the calculated apparent activation energy increases slightly from 147.57 to 161.78 kJ·mol−1. The results of X-ray photoelectron spectroscopy (XPS), Fourier transformation infrared (FT-IR), Raman spectroscopy, and 29Si solid-state magic angular spinning nuclear magnetic resonance (MAS-NMR) reveal that the degree of polymerization (DOP) of slag increases with decreasing the slag basicity, which is consistent with the viscosity variation with basicity. Therefore, the dissolution of silica modifier would be deteriorated as silica content in slag increases. In addition, the 31P solid-state MAS-NMR demonstrates that phosphorous exists in isolated orthophosphate (Q0(P)) structural unit. It was shown that 29Si solid-state MAS-NMR is limited in the structural quantification of Fe-bearing silicate glasses due to the paramagnetic effect of iron ions.Abstract
Slag viscosity is a crucial factor affecting the dissolution of silica modifiers in molten basic oxygen furnace (BOF) slag, particularly during the online modification process. In this work, the viscosity of low basicity CaO-SiO2–10 wt.% Fe2O3–10 wt.% P2O5 slag was measured by the rotating cylinder method. The structure of glassy slag samples was analyzed by multiple spectroscopic techniques to offer a better understanding of the relationship between slag viscosity and structure. It is concluded that the slag viscosity increases with decreasing binary basicity (CaO wt.%/SiO2 wt.%) from 1.46 to 1.16, and the calculated apparent activation energy increases slightly from 147.57 to 161.78 kJ·mol−1. The results of X-ray photoelectron spectroscopy (XPS), Fourier transformation infrared (FT-IR), Raman spectroscopy, and 29Si solid-state magic angular spinning nuclear magnetic resonance (MAS-NMR) reveal that the degree of polymerization (DOP) of slag increases with decreasing the slag basicity, which is consistent with the viscosity variation with basicity. Therefore, the dissolution of silica modifier would be deteriorated as silica content in slag increases. In addition, the 31P solid-state MAS-NMR demonstrates that phosphorous exists in isolated orthophosphate (Q0(P)) structural unit. It was shown that 29Si solid-state MAS-NMR is limited in the structural quantification of Fe-bearing silicate glasses due to the paramagnetic effect of iron ions
Recovery stress control and prediction in NiTi shape memory alloy wires under electrical actuation
No full textAbstract
This study investigates recovery stress control in NiTi SMA wires and utilizes a long short-term memory (LSTM) neural network to predict their recovery behavior. The effects of pre-strain, electrical actuation power, and thermal cycling on recovery stress during active Joule heating are systematically investigated, providing detailed insights under high-cycle operation. The results demonstrate that the pulse width modulation (PWM) driving current frequency must be sufficiently high (≥1000 Hz) to minimize temperature fluctuations. At a fixed maximum heating temperature of 60 °C, recovery stress increases during heating and decreases during cooling, exhibiting a hysteresis effect. The maximum recovery stress at 60 °C depends on pre-strain, increasing initially before plateauing beyond a critical threshold. Under continuous heating, recovery stress rises to a peak at a specific temperature and then gradually decreases. Peak recovery stress increases with pre-strain, and the associated peak temperature shifts upward. The LSTM network with a low loss value of 3.0966 × 10⁻⁴ accurately predicts recovery stress-temperature relationships during heating and cooling. Long-term stability tests confirm that recovery stress remains consistent over 10,000 cycles when actuated with optimal parameters (0.7 W power and 4 % pre-strain). These parameters ensure effective phase transformation while minimizing irreversible plastic deformation. The results highlight the critical role of pre-strain and actuation power in optimizing recovery stress. This insight advances the design of SMA-driven microrobots capable of safely and precisely manipulating fragile objects.Abstract
This study investigates recovery stress control in NiTi SMA wires and utilizes a long short-term memory (LSTM) neural network to predict their recovery behavior. The effects of pre-strain, electrical actuation power, and thermal cycling on recovery stress during active Joule heating are systematically investigated, providing detailed insights under high-cycle operation. The results demonstrate that the pulse width modulation (PWM) driving current frequency must be sufficiently high (≥1000 Hz) to minimize temperature fluctuations. At a fixed maximum heating temperature of 60 °C, recovery stress increases during heating and decreases during cooling, exhibiting a hysteresis effect. The maximum recovery stress at 60 °C depends on pre-strain, increasing initially before plateauing beyond a critical threshold. Under continuous heating, recovery stress rises to a peak at a specific temperature and then gradually decreases. Peak recovery stress increases with pre-strain, and the associated peak temperature shifts upward. The LSTM network with a low loss value of 3.0966 × 10⁻⁴ accurately predicts recovery stress-temperature relationships during heating and cooling. Long-term stability tests confirm that recovery stress remains consistent over 10,000 cycles when actuated with optimal parameters (0.7 W power and 4 % pre-strain). These parameters ensure effective phase transformation while minimizing irreversible plastic deformation. The results highlight the critical role of pre-strain and actuation power in optimizing recovery stress. This insight advances the design of SMA-driven microrobots capable of safely and precisely manipulating fragile objects
Fine-scale patterns and drivers of ploidy state in lentic and lotic macrophyte assemblages across the world
No full textAbstract
To investigate whether patterns of ploidy state variation known to occur in macrophytes at broad global scales can be detected at finer site scale, we examined macrophyte assemblages present in 1239 individual inland lentic and lotic waterbodies sampled from 2000 onwards. The sites include lakes and reservoirs, rivers and streams, slow-flowing or static water bodies associated with rivers (such as oxbows), man-made channels, and temporary or ephemeral lentic waterbodies in 22 countries worldwide. The latitude range for these sites was 10.58–68.40° N and from 0.01 to 54.88° S, covering climatic conditions ranging from tropical to temperate/Arctic. We examined the influence of geospatial variables, current or historic climate variables, and additional local water physical and chemical variables measured for each site, as potential predictors of the incidence of ploidy state (diploidy, polyploidy, and mixed-cytotype) in the macrophyte species assemblage. At fine scales (individual sites), we observed the same latitudinal and climatic patterns influencing all macrophyte ploidy states, especially diploid species, compared to findings at a broad spatial resolution of 10° × 10° latitude-longitude. Ploidy state of macrophyte assemblages slightly, but significantly, differs between lentic and lotic environments. Along with geospatial and climate variables, local physical and chemical variables also helped predict the occurrence of polyploid and mixed-ploidy species. Our results support previous findings on ploidy state distribution and drivers at broader scales but also unravel new information on key drivers for the distribution of polyploid and mixed-ploidy species, relevant to understanding macrophyte adaptation mechanisms and evolutionary processes worldwide.Abstract
To investigate whether patterns of ploidy state variation known to occur in macrophytes at broad global scales can be detected at finer site scale, we examined macrophyte assemblages present in 1239 individual inland lentic and lotic waterbodies sampled from 2000 onwards. The sites include lakes and reservoirs, rivers and streams, slow-flowing or static water bodies associated with rivers (such as oxbows), man-made channels, and temporary or ephemeral lentic waterbodies in 22 countries worldwide. The latitude range for these sites was 10.58–68.40° N and from 0.01 to 54.88° S, covering climatic conditions ranging from tropical to temperate/Arctic. We examined the influence of geospatial variables, current or historic climate variables, and additional local water physical and chemical variables measured for each site, as potential predictors of the incidence of ploidy state (diploidy, polyploidy, and mixed-cytotype) in the macrophyte species assemblage. At fine scales (individual sites), we observed the same latitudinal and climatic patterns influencing all macrophyte ploidy states, especially diploid species, compared to findings at a broad spatial resolution of 10° × 10° latitude-longitude. Ploidy state of macrophyte assemblages slightly, but significantly, differs between lentic and lotic environments. Along with geospatial and climate variables, local physical and chemical variables also helped predict the occurrence of polyploid and mixed-ploidy species. Our results support previous findings on ploidy state distribution and drivers at broader scales but also unravel new information on key drivers for the distribution of polyploid and mixed-ploidy species, relevant to understanding macrophyte adaptation mechanisms and evolutionary processes worldwide
Real-time discharge curve and state of charge estimation of lithium-ion batteries via a physics-informed full homogenized macroscale model
No full textAbstract
Lithium-ion batteries are essential for applications ranging from portable electronics to electric vehicles, playing a key role in addressing climate change. Accurate monitoring of their internal dynamics, such as discharge curves and state of charge (SOC), is crucial for optimizing performance, safety, and lifespan. This study introduces a novel physics-informed neural network (PINN) framework utilizing the full homogenized macroscale (FHM) model to predict key electrochemical parameters under varying loads and temperatures. By embedding physical laws into the network architecture, the framework ensures physically consistent and reliable estimations. The proposed PINN method demonstrates exceptional accuracy, achieving a root mean square error (RMSE) of 0.0389 and an R2 value exceeding 0.97 across all cases. For discharge curves, the mean absolute error (MAE) of discharge voltage remains below 0.02 V for most conditions, while SOC estimation predominantly maintains a MAE under 2 %. These results underscore the framework's robustness, computational efficiency, and suitability for real-time applications, such as battery management systems. Unlike traditional machine learning approaches, which often struggle with highly nonlinear battery behavior, the PINN framework excels in capturing complex electrochemical patterns under high-rate and variable-temperature conditions. Additionally, by leveraging the FHM model, this method balances computational efficiency and accuracy, making it highly applicable to real-world scenarios. This study highlights the transformative potential of physics-informed approaches to advance lithium-ion battery modeling, offering a pathway toward safer, more reliable, and efficient energy storage solutions for diverse applications.Abstract
Lithium-ion batteries are essential for applications ranging from portable electronics to electric vehicles, playing a key role in addressing climate change. Accurate monitoring of their internal dynamics, such as discharge curves and state of charge (SOC), is crucial for optimizing performance, safety, and lifespan. This study introduces a novel physics-informed neural network (PINN) framework utilizing the full homogenized macroscale (FHM) model to predict key electrochemical parameters under varying loads and temperatures. By embedding physical laws into the network architecture, the framework ensures physically consistent and reliable estimations. The proposed PINN method demonstrates exceptional accuracy, achieving a root mean square error (RMSE) of 0.0389 and an R2 value exceeding 0.97 across all cases. For discharge curves, the mean absolute error (MAE) of discharge voltage remains below 0.02 V for most conditions, while SOC estimation predominantly maintains a MAE under 2 %. These results underscore the framework's robustness, computational efficiency, and suitability for real-time applications, such as battery management systems. Unlike traditional machine learning approaches, which often struggle with highly nonlinear battery behavior, the PINN framework excels in capturing complex electrochemical patterns under high-rate and variable-temperature conditions. Additionally, by leveraging the FHM model, this method balances computational efficiency and accuracy, making it highly applicable to real-world scenarios. This study highlights the transformative potential of physics-informed approaches to advance lithium-ion battery modeling, offering a pathway toward safer, more reliable, and efficient energy storage solutions for diverse applications
SRST: A secure and resilient synchronization of time for WSNs in IoT applications
No full textAbstract
Many applications of Wireless Sensor Networks (WSNs) in internet of things require accurate time synchronization for the successful realization. In a hostile environment, protecting time synchronization against faulty timestamps injection attacks is paramount. Malicious nodes (MNs) could broadcast faulty timestamps to decrease the accuracy of the whole network. The Receiver-Only (RO)-based synchronization methodology achieves high accuracy while reducing the number of timing messages compared to other methodologies. However, RO is vulnerable to node failure and supports only a single reference, which makes it inaccurate in the presence of MNs. To address these limitations, we propose SRST, a secure and resilient synchronization of time for WSNs. SRST employs a novel time synchronization model that extends RO to allow synchronizing sensor nodes to multiple mutually-synchronized references, which enhances robustness against node failure and malicious behavior. SRST optimizes convergence time by synchronizing the RO node with its synchronized 1-hop and 2-hop neighbors. RO node applies a new delay threshold-based detection technique to identify reference nodes as MNs through detection of faulty timestamps. SRST is topology-independent and can be applied to multi-tier, cluster-based and flat topologies. We validated SRST through simulations and prototype experiments, comparing its performance with several state-of-the-art protocols. The results demonstrate that SRST outperforms existing protocols in accuracy, achieving synchronization within less than , and accelerates convergence time by a factor of 31.25 compared to the best-known protocol. SRST has also been shown to be more effective in mitigating faulty timestamp injection attacks, successfully detecting errors of with a 100% success rate and ensuring that the clocks of any two nodes do not deviate by more than . Furthermore, the results indicate that SRST imposes little communication overhead.Abstract
Many applications of Wireless Sensor Networks (WSNs) in internet of things require accurate time synchronization for the successful realization. In a hostile environment, protecting time synchronization against faulty timestamps injection attacks is paramount. Malicious nodes (MNs) could broadcast faulty timestamps to decrease the accuracy of the whole network. The Receiver-Only (RO)-based synchronization methodology achieves high accuracy while reducing the number of timing messages compared to other methodologies. However, RO is vulnerable to node failure and supports only a single reference, which makes it inaccurate in the presence of MNs. To address these limitations, we propose SRST, a secure and resilient synchronization of time for WSNs. SRST employs a novel time synchronization model that extends RO to allow synchronizing sensor nodes to multiple mutually-synchronized references, which enhances robustness against node failure and malicious behavior. SRST optimizes convergence time by synchronizing the RO node with its synchronized 1-hop and 2-hop neighbors. RO node applies a new delay threshold-based detection technique to identify reference nodes as MNs through detection of faulty timestamps. SRST is topology-independent and can be applied to multi-tier, cluster-based and flat topologies. We validated SRST through simulations and prototype experiments, comparing its performance with several state-of-the-art protocols. The results demonstrate that SRST outperforms existing protocols in accuracy, achieving synchronization within less than , and accelerates convergence time by a factor of 31.25 compared to the best-known protocol. SRST has also been shown to be more effective in mitigating faulty timestamp injection attacks, successfully detecting errors of with a 100% success rate and ensuring that the clocks of any two nodes do not deviate by more than . Furthermore, the results indicate that SRST imposes little communication overhead