Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652929
Erik J. Jorgensen, Frank Werner, Milos Prvulović, A. Zajić
Electromagnetic (EM) side-channel radiation visualized with a spectrogram can be used to classify program states of a computer processor. However, clustering a spectrogram to automatically track program states is difficult due to their often noisy nature. Popular clustering algorithms like K-Means or HDBSCAN fail to adequately cluster spectrogram samples into the variable-length subsequences that define the program states. These algorithms do not account for the time-continuity of spectrogram samples and consequently tend to assign spurious cluster label changes between samples. Here we develop an algorithm, called Hierarchical Subsequence Clustering for Spectrograms, that uses an intuitive approach to explicitly constrain the clustering problem and generate time-continuous clusters. We demonstrate through experiments with simulated program activity as well as with real EM side-channel data measured from a running cellphone that our automated clustering method is faster and yields better clusters in the presence of significant noise.
{"title":"A Hierarchical Subsequence Clustering Method for Tracking Program States in Spectrograms","authors":"Erik J. Jorgensen, Frank Werner, Milos Prvulović, A. Zajić","doi":"10.1109/MILCOM52596.2021.9652929","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652929","url":null,"abstract":"Electromagnetic (EM) side-channel radiation visualized with a spectrogram can be used to classify program states of a computer processor. However, clustering a spectrogram to automatically track program states is difficult due to their often noisy nature. Popular clustering algorithms like K-Means or HDBSCAN fail to adequately cluster spectrogram samples into the variable-length subsequences that define the program states. These algorithms do not account for the time-continuity of spectrogram samples and consequently tend to assign spurious cluster label changes between samples. Here we develop an algorithm, called Hierarchical Subsequence Clustering for Spectrograms, that uses an intuitive approach to explicitly constrain the clustering problem and generate time-continuous clusters. We demonstrate through experiments with simulated program activity as well as with real EM side-channel data measured from a running cellphone that our automated clustering method is faster and yields better clusters in the presence of significant noise.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"196 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133576264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652942
Daniel Green, M. Tummala, J. McEachen
This paper explores the use of wavelet analysis and deep learning techniques to classify pulsed binary data over a severely noisy communications channel. Military communications need to operate in extremely harsh radio environments, which may include hostile intent to disrupt communications. Consequently, unconventional methods, such as pulsed communications, need to be investigated. Pulsed transmission techniques utilized for such channels typically result in pulses that are not easily discerned from noise and other interference. Deep learning techniques have proven advantageous in quickly and efficiently identifying minute variations in large data sets. This paper presents methods for leveraging deep learning techniques for pulsed signal detection.
{"title":"Pulsed Signal Detection Utilizing Wavelet Analysis with a Deep Learning Approach","authors":"Daniel Green, M. Tummala, J. McEachen","doi":"10.1109/MILCOM52596.2021.9652942","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652942","url":null,"abstract":"This paper explores the use of wavelet analysis and deep learning techniques to classify pulsed binary data over a severely noisy communications channel. Military communications need to operate in extremely harsh radio environments, which may include hostile intent to disrupt communications. Consequently, unconventional methods, such as pulsed communications, need to be investigated. Pulsed transmission techniques utilized for such channels typically result in pulses that are not easily discerned from noise and other interference. Deep learning techniques have proven advantageous in quickly and efficiently identifying minute variations in large data sets. This paper presents methods for leveraging deep learning techniques for pulsed signal detection.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133668379","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9653094
M. Vai, David Whelihan, K. Denney, Robert Lychev, Jeffrey J. Hughes, Donato Kava, Alice Lee, Nicholas Evancich, Richard Clark, D. Lide, K. Kwak, Jason H. Li, Douglas Schafer, Michael Lynch, Kyle Tillotson, Wladimir Tirenin
Mission assurance requires all operational platforms and systems to be able to perform their function under the ever-growing challenge of cyber hostility. Since 2014, the Air Force Research Laboratory (AFRL) Agile and Resilient Embedded System (ARES) program has been developing a Cyber Security and Resilience (CSR) methodology and associated technologies for ground-up designs as well as retrofitting existing platforms. Throughout the duration of the program, the ARES approach was applied and demonstrated with an octocopter, an unmanned aerial system testbed, and a fixed-wing aircraft. Since then, efforts have been focused on advancing capability and transitioning CSR technologies into DoD systems. In this paper, we summarize the ARES methodology and technologies, and describe our experience of inserting them into DoD embedded systems.
{"title":"Agile and Resilient Embedded Systems","authors":"M. Vai, David Whelihan, K. Denney, Robert Lychev, Jeffrey J. Hughes, Donato Kava, Alice Lee, Nicholas Evancich, Richard Clark, D. Lide, K. Kwak, Jason H. Li, Douglas Schafer, Michael Lynch, Kyle Tillotson, Wladimir Tirenin","doi":"10.1109/MILCOM52596.2021.9653094","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9653094","url":null,"abstract":"Mission assurance requires all operational platforms and systems to be able to perform their function under the ever-growing challenge of cyber hostility. Since 2014, the Air Force Research Laboratory (AFRL) Agile and Resilient Embedded System (ARES) program has been developing a Cyber Security and Resilience (CSR) methodology and associated technologies for ground-up designs as well as retrofitting existing platforms. Throughout the duration of the program, the ARES approach was applied and demonstrated with an octocopter, an unmanned aerial system testbed, and a fixed-wing aircraft. Since then, efforts have been focused on advancing capability and transitioning CSR technologies into DoD systems. In this paper, we summarize the ARES methodology and technologies, and describe our experience of inserting them into DoD embedded systems.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116787955","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9653069
Nathaniel B. Soule, Brandon Kalashian, Colleen T. Rock, Landon Tomcho
From disaster relief to combat search and rescue, mobile devices are increasingly key to mission success at the tactical edge. These mobile devices typically rely on networked communications to fulfil some of their most important functions. Unfortunately, due to cost, complexity, internal processes or other factors in both military and civilian scenarios, Internet Protocol (IP) networks are not always available to support such communications. Analog radios, in the form of everything from general purpose COTS walkie talkies to DoD tactical handhelds, are often all that is accessible. Today's mobile devices and applications, such as the Android Tactical Assault Kit (ATAK) – a phone-based situational awareness tool, cannot send their digital data directly over analog signals, however, and thus historically have been unable to capitalize on this large set of prevalent and often affordable radios around them. The Handheld Acoustic Modem for Mobile Exchanges with Radios (HAMMER) ATAK plugin is a software acoustic modem that allows ATAK devices to communicate with each other using any voice-comms capable radio, without the need for additional hardware. This paper describes the HAMMER technology, its military and civilian applications, current challenges and constraints, and evaluates the tool in several contexts.
{"title":"Software Acoustic Modem for TAK Communications with Analog Radios at the Tactical Edge","authors":"Nathaniel B. Soule, Brandon Kalashian, Colleen T. Rock, Landon Tomcho","doi":"10.1109/MILCOM52596.2021.9653069","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9653069","url":null,"abstract":"From disaster relief to combat search and rescue, mobile devices are increasingly key to mission success at the tactical edge. These mobile devices typically rely on networked communications to fulfil some of their most important functions. Unfortunately, due to cost, complexity, internal processes or other factors in both military and civilian scenarios, Internet Protocol (IP) networks are not always available to support such communications. Analog radios, in the form of everything from general purpose COTS walkie talkies to DoD tactical handhelds, are often all that is accessible. Today's mobile devices and applications, such as the Android Tactical Assault Kit (ATAK) – a phone-based situational awareness tool, cannot send their digital data directly over analog signals, however, and thus historically have been unable to capitalize on this large set of prevalent and often affordable radios around them. The Handheld Acoustic Modem for Mobile Exchanges with Radios (HAMMER) ATAK plugin is a software acoustic modem that allows ATAK devices to communicate with each other using any voice-comms capable radio, without the need for additional hardware. This paper describes the HAMMER technology, its military and civilian applications, current challenges and constraints, and evaluates the tool in several contexts.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125825111","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652922
Zheyu Chen, K. Leung, Shiqiang Wang, L. Tassiulas, Kevin S. Chan
Many technical issues for communications and computer infrastructures, including resource sharing, network management and distributed analytics, can be formulated as optimization problems. Gradient-based iterative algorithms have been widely utilized to solve these problems. Much research focuses on improving the iteration convergence. However, when system parameters change, it requires a new solution from the iterative methods. Therefore, it is helpful to develop machine-learning solution frameworks that can quickly produce solutions over a range of system parameters. We propose here a learning approach to solve non-convex, constrained optimization problems. Two coupled Long Short Term Memory (LSTM) networks are used to find the optimal solution. The advantages of this new framework include: (1) near optimal solution for a given problem instance can be obtained in very few iterations (time steps) during the inference process, (2) the learning approach allows selections of various hyper-parameters to achieve desirable tradeoffs between the training time and the solution quality, and (3) the coupled-LSTM networks can be trained using system parameters with distributions different from those used during inference to generate solutions, thus enhancing the robustness of the learning technique. Numerical experiments using a dataset from Alibaba reveal that the relative discrepancy between the generated solution and the optimum is less than 1% and 0.1% after 2 and 12 iterations, respectively.
{"title":"Robust Solutions to Constrained Optimization Problems by LSTM Networks","authors":"Zheyu Chen, K. Leung, Shiqiang Wang, L. Tassiulas, Kevin S. Chan","doi":"10.1109/MILCOM52596.2021.9652922","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652922","url":null,"abstract":"Many technical issues for communications and computer infrastructures, including resource sharing, network management and distributed analytics, can be formulated as optimization problems. Gradient-based iterative algorithms have been widely utilized to solve these problems. Much research focuses on improving the iteration convergence. However, when system parameters change, it requires a new solution from the iterative methods. Therefore, it is helpful to develop machine-learning solution frameworks that can quickly produce solutions over a range of system parameters. We propose here a learning approach to solve non-convex, constrained optimization problems. Two coupled Long Short Term Memory (LSTM) networks are used to find the optimal solution. The advantages of this new framework include: (1) near optimal solution for a given problem instance can be obtained in very few iterations (time steps) during the inference process, (2) the learning approach allows selections of various hyper-parameters to achieve desirable tradeoffs between the training time and the solution quality, and (3) the coupled-LSTM networks can be trained using system parameters with distributions different from those used during inference to generate solutions, thus enhancing the robustness of the learning technique. Numerical experiments using a dataset from Alibaba reveal that the relative discrepancy between the generated solution and the optimum is less than 1% and 0.1% after 2 and 12 iterations, respectively.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121972287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9653065
Mengkun Ji, K. Chugg
This paper introduces two approaches to compute the average sampling frequency (ASF) of ideal level crossing analog-to-digital converters (LC-ADCs). The first is based on Rice's analysis method and can be used in various combinations of Gaussian signals. The second, a direct method, can only be used for narrowband modulated sinusoidal carrier input signals. These analysis results agree very well with computer simulations for ideal LC-ADCs and also highlight the oversampling issue for LC-ADCs (i.e., sampling at rates higher than Nyquist). Wu and Chen previously proposed a Gated LC-ADC to address this oversampling issue. We develop an approximate analysis for the ASF of this Gated LC-ADC by modeling the samples from the un-Gated LC-ADC as a Poisson arrival process. This approximation captures the desired effect of eliminating the oversampling issue reasonably well.
{"title":"Analysis of the Average Sampling Frequency for Level Crossing Analog-to-Digital Converters","authors":"Mengkun Ji, K. Chugg","doi":"10.1109/MILCOM52596.2021.9653065","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9653065","url":null,"abstract":"This paper introduces two approaches to compute the average sampling frequency (ASF) of ideal level crossing analog-to-digital converters (LC-ADCs). The first is based on Rice's analysis method and can be used in various combinations of Gaussian signals. The second, a direct method, can only be used for narrowband modulated sinusoidal carrier input signals. These analysis results agree very well with computer simulations for ideal LC-ADCs and also highlight the oversampling issue for LC-ADCs (i.e., sampling at rates higher than Nyquist). Wu and Chen previously proposed a Gated LC-ADC to address this oversampling issue. We develop an approximate analysis for the ASF of this Gated LC-ADC by modeling the samples from the un-Gated LC-ADC as a Poisson arrival process. This approximation captures the desired effect of eliminating the oversampling issue reasonably well.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130352931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652943
S. Sruti, Chilaka Deepti, K. Giridhar
Distributed MIMO radar systems offer tremendous advantage in the detection of airborne platforms employing stealth and are resilient to single point failure. However, when multiple targets are present over the surveillance region, the reflected signals at various receivers from these targets cannot be uniquely associated to the targets easily. Incorrect associations of the received data lead to the creation of ghost targets, and hence, de-ghosting is an inherent problem in distributed radar systems. Exploiting the geometry of the measurement model into the association process, we devise algorithms that are practically implementable and computationally feasible. In this work, a novel, efficient and fast data association scheme followed by a localization algorithm is proposed that utilizes Time-of-Arrival and Doppler frequency measurements of the targets with respect to the transmitter-receiver pairs to accurately determine 3D position and velocities of the targets. The proposed approach is non-parametric as it does not need the assumption of initial states, number of targets and their motion models. It simultaneously associates up to four targets present within a minimum horizontal separation of $100mtimes 100m$ for signals of bandwidth 20MHz and any number of targets that are flying far away from this minimum separation in the observation region. It can also associate and track up to nine targets that have sequential birth and random death, flying with random realizable velocities.
{"title":"Non-Parametric and Geometric Multi-Target Data Association for Distributed MIMO Radars","authors":"S. Sruti, Chilaka Deepti, K. Giridhar","doi":"10.1109/MILCOM52596.2021.9652943","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652943","url":null,"abstract":"Distributed MIMO radar systems offer tremendous advantage in the detection of airborne platforms employing stealth and are resilient to single point failure. However, when multiple targets are present over the surveillance region, the reflected signals at various receivers from these targets cannot be uniquely associated to the targets easily. Incorrect associations of the received data lead to the creation of ghost targets, and hence, de-ghosting is an inherent problem in distributed radar systems. Exploiting the geometry of the measurement model into the association process, we devise algorithms that are practically implementable and computationally feasible. In this work, a novel, efficient and fast data association scheme followed by a localization algorithm is proposed that utilizes Time-of-Arrival and Doppler frequency measurements of the targets with respect to the transmitter-receiver pairs to accurately determine 3D position and velocities of the targets. The proposed approach is non-parametric as it does not need the assumption of initial states, number of targets and their motion models. It simultaneously associates up to four targets present within a minimum horizontal separation of $100mtimes 100m$ for signals of bandwidth 20MHz and any number of targets that are flying far away from this minimum separation in the observation region. It can also associate and track up to nine targets that have sequential birth and random death, flying with random realizable velocities.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127137931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652983
Brandon T. Hunt, David B. Haab, Thomas Cameron Sego, Tom V. Holschuh, H. Moradi, B. Farhang-Boroujeny
Filter bank multicarrier spread spectrum (FBMC-SS) has proven to be a robust and reliable waveform choice for communication over high frequency (HF) skywave links. However, the performance of this waveform has yet to be contextualized against typical robust HF waveforms, such as the Walsh-encoded waveform detailed in the MIL-STD-188-110D, Appendix D document. In this paper, we first outline the advantages of both the Walsh and FBMC-SS waveforms as well as present their developments. Simulation results are then presented for ideal, simulated HF, and HF with interference channel conditions. Lastly, skywave-HF results are presented for these two waveforms both with and without interference.
{"title":"Examining the Performance of Walsh-DSSS Against FBMC-SS in HF Channels","authors":"Brandon T. Hunt, David B. Haab, Thomas Cameron Sego, Tom V. Holschuh, H. Moradi, B. Farhang-Boroujeny","doi":"10.1109/MILCOM52596.2021.9652983","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652983","url":null,"abstract":"Filter bank multicarrier spread spectrum (FBMC-SS) has proven to be a robust and reliable waveform choice for communication over high frequency (HF) skywave links. However, the performance of this waveform has yet to be contextualized against typical robust HF waveforms, such as the Walsh-encoded waveform detailed in the MIL-STD-188-110D, Appendix D document. In this paper, we first outline the advantages of both the Walsh and FBMC-SS waveforms as well as present their developments. Simulation results are then presented for ideal, simulated HF, and HF with interference channel conditions. Lastly, skywave-HF results are presented for these two waveforms both with and without interference.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128891613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9652959
Kevin Talty, J. Stockdale, Nathaniel D. Bastian
As the demand for data has increased, we have witnessed a surge in the use of machine learning to help aid industry and government in making sense of massive amounts of data and, subsequently, making predictions and decisions. For the military, this surge has manifested itself in the Internet of Battlefield Things. The pervasive nature of data on today's battlefield will allow machine learning models to increase soldier lethality and survivability. However, machine learning models are predicated upon the assumptions that the data upon which these machine learning models are being trained is truthful and the machine learning models are not compromised. These assumptions surrounding the quality of data and models cannot be the status-quo going forward as attackers establish novel methods to exploit machine learning models for their benefit. These novel attack methods can be described as adversarial machine learning (AML). These attacks allow an attacker to unsuspectingly alter a machine learning model before and after model training in order to degrade a model's ability to detect malicious activity. In this paper, we show how AML, by poisoning data sets and evading well trained models, affect machine learning models' ability to function as Network Intrusion Detection Systems (NIDS). Finally, we highlight why evasion attacks are especially effective in this setting and discuss some of the causes for this degradation of model effectiveness.
{"title":"A Sensitivity Analysis of Poisoning and Evasion Attacks in Network Intrusion Detection System Machine Learning Models","authors":"Kevin Talty, J. Stockdale, Nathaniel D. Bastian","doi":"10.1109/MILCOM52596.2021.9652959","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9652959","url":null,"abstract":"As the demand for data has increased, we have witnessed a surge in the use of machine learning to help aid industry and government in making sense of massive amounts of data and, subsequently, making predictions and decisions. For the military, this surge has manifested itself in the Internet of Battlefield Things. The pervasive nature of data on today's battlefield will allow machine learning models to increase soldier lethality and survivability. However, machine learning models are predicated upon the assumptions that the data upon which these machine learning models are being trained is truthful and the machine learning models are not compromised. These assumptions surrounding the quality of data and models cannot be the status-quo going forward as attackers establish novel methods to exploit machine learning models for their benefit. These novel attack methods can be described as adversarial machine learning (AML). These attacks allow an attacker to unsuspectingly alter a machine learning model before and after model training in order to degrade a model's ability to detect malicious activity. In this paper, we show how AML, by poisoning data sets and evading well trained models, affect machine learning models' ability to function as Network Intrusion Detection Systems (NIDS). Finally, we highlight why evasion attacks are especially effective in this setting and discuss some of the causes for this degradation of model effectiveness.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"108 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127681333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-11-29DOI: 10.1109/MILCOM52596.2021.9653127
David Nieves-Acaron, Benjamin Luchterhand, A. Aravamudan, David Elliott, Steven Wyatt, Carlos E. Otero, L. D. Otero, Anthony O. Smith, A. Peter, Wesley Jones, Eric Lam
Superior battlefield Situational Awareness (SA) requires timely and coherent integration of various sensor modalities to provide the most complete, real-time picture of in-theater activities. In this work, we introduce Acoustic Classification at the Edge (ACE), an ATAK plugin for improved acoustic SA, to move beyond traditional full-motion video and geospatial data typically employed for SA, and instead focus on acoustic intelligence. Our Android Tactical Awareness Kit (ATAK) plugin is able to perform on-device audio recording, classification, labeling, and autonomous reach-back to the cloud, when available, to enable warfighters to improve SA over time. As part of ACE, we detail a machine learning analytic designed to classify acoustic sources directly at the edge, with a case study on firearm classification. We also detail the cloud infrastructure necessary to support it. This paper describes the application and cloud architecture, in-theater operations, and experimental results after having deployed the plugin on ATAK. Finally, we propose future directions for acoustic classification at the edge based on our findings.
{"title":"ACE: An ATAK Plugin for Enhanced Acoustic Situational Awareness at the Edge","authors":"David Nieves-Acaron, Benjamin Luchterhand, A. Aravamudan, David Elliott, Steven Wyatt, Carlos E. Otero, L. D. Otero, Anthony O. Smith, A. Peter, Wesley Jones, Eric Lam","doi":"10.1109/MILCOM52596.2021.9653127","DOIUrl":"https://doi.org/10.1109/MILCOM52596.2021.9653127","url":null,"abstract":"Superior battlefield Situational Awareness (SA) requires timely and coherent integration of various sensor modalities to provide the most complete, real-time picture of in-theater activities. In this work, we introduce Acoustic Classification at the Edge (ACE), an ATAK plugin for improved acoustic SA, to move beyond traditional full-motion video and geospatial data typically employed for SA, and instead focus on acoustic intelligence. Our Android Tactical Awareness Kit (ATAK) plugin is able to perform on-device audio recording, classification, labeling, and autonomous reach-back to the cloud, when available, to enable warfighters to improve SA over time. As part of ACE, we detail a machine learning analytic designed to classify acoustic sources directly at the edge, with a case study on firearm classification. We also detail the cloud infrastructure necessary to support it. This paper describes the application and cloud architecture, in-theater operations, and experimental results after having deployed the plugin on ATAK. Finally, we propose future directions for acoustic classification at the edge based on our findings.","PeriodicalId":187645,"journal":{"name":"MILCOM 2021 - 2021 IEEE Military Communications Conference (MILCOM)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130116414","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：