Pub Date : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555416
Faqeer ur Rehman, C. Izurieta
Enhancing the trust of machine learning-based classifiers with large input spaces is a desirable goal; however, due to high labeling costs and limited resources, this is a challenging task. One solution is to use test input prioritization techniques that aim to identify and label only the most effective test instances. These prioritized test inputs can then be used with some popular testing techniques e.g., Metamorphic testing (MT) to test and uncover implementation bugs in computationally complex machine learning classifiers that suffer from the oracle problem. However, there are certain limitations involved with this approach, (i) using a small number of prioritized test inputs may not be enough to check the program correctness over a large variety of input scenarios, and (ii) traditional MT approaches become infeasible when the programs under test exhibit a non-deterministic behavior during training e.g., Neural Network-based classifiers. Therefore, instead of using MT for testing purposes, we propose a metamorphic relation to solve a data generation/labeling problem; that is, enhancing the test inputs effectiveness by extending the prioritized test set with new tests without incurring additional labeling costs. Further, we leverage the prioritized test inputs (both source and follow-up data sets) and propose a statistical hypothesis testing (for detection) and machine learning-based approach (for prediction) of faulty behavior in two other machine learning classifiers (Neural Network-based Intrusion Detection Systems). In our case, the problem is interesting in the sense that injected bugs represent the high accuracy producing mutated program versions that may be difficult to detect by a software developer. The results indicate that (i) the proposed statistical hypothesis testing is able to identify the induced buggy behavior, and (ii) Random Forest outperforms and achieves the best performance over SVM and k-NN algorithms.
增强具有大输入空间的基于机器学习的分类器的信任是一个理想的目标;然而，由于高标签成本和有限的资源，这是一项具有挑战性的任务。一种解决方案是使用测试输入优先级技术，旨在识别和标记最有效的测试实例。然后，这些优先级的测试输入可以与一些流行的测试技术一起使用，例如，变形测试(MT)，以测试和发现计算复杂的机器学习分类器中遭受oracle问题的实现错误。然而，这种方法有一定的局限性，(i)使用少量的优先测试输入可能不足以检查程序在各种输入场景下的正确性，(ii)当被测试程序在训练期间表现出不确定性行为时，传统的机器翻译方法变得不可行的，例如基于神经网络的分类器。因此，我们提出了一种变质关系来解决数据生成/标记问题，而不是使用MT进行测试;也就是说，通过使用新测试扩展优先测试集来增强测试输入的有效性，而不会产生额外的标记成本。此外，我们利用了优先测试输入(源数据集和后续数据集)，并在另外两个机器学习分类器(基于神经网络的入侵检测系统)中提出了错误行为的统计假设检验(用于检测)和基于机器学习的方法(用于预测)。在我们的例子中，这个问题很有趣，因为注入的错误代表了产生突变程序版本的高精度，这可能很难被软件开发人员检测到。结果表明:(i)所提出的统计假设检验能够识别诱导的bug行为，(ii) Random Forest优于SVM和k-NN算法，并取得了最佳性能。
{"title":"A Hybridized Approach for Testing Neural Network Based Intrusion Detection Systems","authors":"Faqeer ur Rehman, C. Izurieta","doi":"10.1109/SmartNets50376.2021.9555416","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555416","url":null,"abstract":"Enhancing the trust of machine learning-based classifiers with large input spaces is a desirable goal; however, due to high labeling costs and limited resources, this is a challenging task. One solution is to use test input prioritization techniques that aim to identify and label only the most effective test instances. These prioritized test inputs can then be used with some popular testing techniques e.g., Metamorphic testing (MT) to test and uncover implementation bugs in computationally complex machine learning classifiers that suffer from the oracle problem. However, there are certain limitations involved with this approach, (i) using a small number of prioritized test inputs may not be enough to check the program correctness over a large variety of input scenarios, and (ii) traditional MT approaches become infeasible when the programs under test exhibit a non-deterministic behavior during training e.g., Neural Network-based classifiers. Therefore, instead of using MT for testing purposes, we propose a metamorphic relation to solve a data generation/labeling problem; that is, enhancing the test inputs effectiveness by extending the prioritized test set with new tests without incurring additional labeling costs. Further, we leverage the prioritized test inputs (both source and follow-up data sets) and propose a statistical hypothesis testing (for detection) and machine learning-based approach (for prediction) of faulty behavior in two other machine learning classifiers (Neural Network-based Intrusion Detection Systems). In our case, the problem is interesting in the sense that injected bugs represent the high accuracy producing mutated program versions that may be difficult to detect by a software developer. The results indicate that (i) the proposed statistical hypothesis testing is able to identify the induced buggy behavior, and (ii) Random Forest outperforms and achieves the best performance over SVM and k-NN algorithms.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"123 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121021439","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 : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555409
Di Wang, Ahmad Al-Rubaie, Yaqoub Alsarkal, Sandra Stincic, John Davies
Automatic meta-data extraction from images from highway cameras is a necessary component for intelligent transportation and smart city. Meta-data can include detailed information on vehicles, such as car make/model, car registration plate and drivers’ behaviour, etc.. This paper focuses on real-time car make/model information extraction from highway cameras. As we have very limited access to the real world data due to data privacy and protection, we use open-source data (e.g. car selling websites) and transfer learning on open-source pre-trained models to build a model which is generic enough to be applied directly to similar data sets from other sources, (e.g. real-world highway cameras) without losing much accuracy. To achieve this, we propose applying the object detection method ‘You Only Look Once’ (Yolo) for classification problem of car make/model. The proposed method and trained model achieve an accuracy of 95.6% when applied directly to real-world highway cameras without using their data for training.
高速公路摄像头图像元数据自动提取是智能交通和智慧城市的必要组成部分。元数据可以包括车辆的详细信息，如汽车的品牌/型号，汽车的车牌和司机的行为等。本文的研究重点是公路摄像头中实时的车型信息提取。由于数据隐私和保护，我们对现实世界数据的访问非常有限，我们使用开源数据(例如汽车销售网站)并在开源预训练模型上进行迁移学习，以构建一个足够通用的模型，可以直接应用于来自其他来源的类似数据集(例如现实世界的高速公路摄像头)，而不会失去太多准确性。为了实现这一目标，我们提出将目标检测方法“You Only Look Once”(Yolo)应用于汽车品牌/型号的分类问题。在不使用真实公路摄像头数据进行训练的情况下，将所提出的方法和训练好的模型直接应用于真实公路摄像头，准确率达到95.6%。
{"title":"Cost effective and Accurate Vehicle Make/Model Recognition method Using YoloV5","authors":"Di Wang, Ahmad Al-Rubaie, Yaqoub Alsarkal, Sandra Stincic, John Davies","doi":"10.1109/SmartNets50376.2021.9555409","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555409","url":null,"abstract":"Automatic meta-data extraction from images from highway cameras is a necessary component for intelligent transportation and smart city. Meta-data can include detailed information on vehicles, such as car make/model, car registration plate and drivers’ behaviour, etc.. This paper focuses on real-time car make/model information extraction from highway cameras. As we have very limited access to the real world data due to data privacy and protection, we use open-source data (e.g. car selling websites) and transfer learning on open-source pre-trained models to build a model which is generic enough to be applied directly to similar data sets from other sources, (e.g. real-world highway cameras) without losing much accuracy. To achieve this, we propose applying the object detection method ‘You Only Look Once’ (Yolo) for classification problem of car make/model. The proposed method and trained model achieve an accuracy of 95.6% when applied directly to real-world highway cameras without using their data for training.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121694562","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 : 2021-09-22DOI: 10.1109/smartnets50376.2021.9555413
{"title":"SmartNets 2021 Authors Index","authors":"","doi":"10.1109/smartnets50376.2021.9555413","DOIUrl":"https://doi.org/10.1109/smartnets50376.2021.9555413","url":null,"abstract":"","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133183210","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 : 2021-09-22DOI: 10.1109/smartnets50376.2021.9555427
{"title":"[SmartNets 2021 Front cover]","authors":"","doi":"10.1109/smartnets50376.2021.9555427","DOIUrl":"https://doi.org/10.1109/smartnets50376.2021.9555427","url":null,"abstract":"","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129800799","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 : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555431
Caolan Deery, Kevin Meehan
The ongoing COVID-19 pandemic has changed people’s lives in ways that many would not have predicted. In the days, weeks and months since mandatory lockdowns and restrictions came into effect worldwide, people have had to adjust their daily lives in an effort to slow and restrict the spread of the virus -- like regularly sanitising their hands, maintaining social distancing in crowded places, and wearing facemasks. The latter is contentious for some but has been a necessary deterrent in slowing the spread of this virus. There is potential for utilising technology as a supplementary deterrent and monitoring tool to help detect non-compliance of mask wearing. This research investigates the efficacy of AI for such purposes, exploring the applicability of a Convolutional Neural Network (CNN), for predicting if a person in a real time video feed is wearing a facemask. A dataset of over 10,000 images was created to effectively evaluate this research. The CNN developed was tested against the validation dataset to evaluate its performance, the model demonstrated 98.47% accuracy on a varied and balanced dataset.
{"title":"Using Deep Learning for COVID-19 Control: Implementing a Convolutional Neural Network in a Facemask Detection Application","authors":"Caolan Deery, Kevin Meehan","doi":"10.1109/SmartNets50376.2021.9555431","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555431","url":null,"abstract":"The ongoing COVID-19 pandemic has changed people’s lives in ways that many would not have predicted. In the days, weeks and months since mandatory lockdowns and restrictions came into effect worldwide, people have had to adjust their daily lives in an effort to slow and restrict the spread of the virus -- like regularly sanitising their hands, maintaining social distancing in crowded places, and wearing facemasks. The latter is contentious for some but has been a necessary deterrent in slowing the spread of this virus. There is potential for utilising technology as a supplementary deterrent and monitoring tool to help detect non-compliance of mask wearing. This research investigates the efficacy of AI for such purposes, exploring the applicability of a Convolutional Neural Network (CNN), for predicting if a person in a real time video feed is wearing a facemask. A dataset of over 10,000 images was created to effectively evaluate this research. The CNN developed was tested against the validation dataset to evaluate its performance, the model demonstrated 98.47% accuracy on a varied and balanced dataset.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121604871","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 : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555429
Sraddhanjali Acharya, Abdul Serwadda
Studies on the characterization of the dexterity of fingers and hands improve the understanding of how humans interact with computing devices. In this study, finger bending patterns captured by flex sensors worn on the fingers are characterized to build a biometric authentication system. The modality uses an array of resistive sensors fitted in a smart glove worn by users while typing. The study encompasses 55 users, 23 of them entered a 9-digit PIN on a laptop’s number pad, and 32 of them typed a 10-length alphanumeric password on the full-sized keyboard. The results demonstrate that the users are authenticated using features built from the flex sensors relating to their PIN and password with a mean EER score of 7.49% and 9.76%, respectively. We further assessed the potential of using individual fingers to authenticate users in both the biometric systems and found that even the fingers not used for typing exhibited discriminative patterns due to movement dynamics during the typing process. This assessment highlights the potential for designing lightweight biometric modalities utilizing dexterity and patterns based on fewer fingers.
{"title":"On Finger Stretching and Bending Dynamics as a Biometric Modality","authors":"Sraddhanjali Acharya, Abdul Serwadda","doi":"10.1109/SmartNets50376.2021.9555429","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555429","url":null,"abstract":"Studies on the characterization of the dexterity of fingers and hands improve the understanding of how humans interact with computing devices. In this study, finger bending patterns captured by flex sensors worn on the fingers are characterized to build a biometric authentication system. The modality uses an array of resistive sensors fitted in a smart glove worn by users while typing. The study encompasses 55 users, 23 of them entered a 9-digit PIN on a laptop’s number pad, and 32 of them typed a 10-length alphanumeric password on the full-sized keyboard. The results demonstrate that the users are authenticated using features built from the flex sensors relating to their PIN and password with a mean EER score of 7.49% and 9.76%, respectively. We further assessed the potential of using individual fingers to authenticate users in both the biometric systems and found that even the fingers not used for typing exhibited discriminative patterns due to movement dynamics during the typing process. This assessment highlights the potential for designing lightweight biometric modalities utilizing dexterity and patterns based on fewer fingers.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122769817","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}
Existing works on mobile edge computing (MEC) which operate under multi-user and multi-server scenarios often assume centralized computation offloading. This paper proposes a decentralized computation offloading scheme where a user does not require information about other users and about the MEC network (e.g., number of servers, network topology). Under the proposed computation offloading scheme, a user infers the transmission delay from the link rate assigned by its associated base station (BS). Further, each user privately deploys a moving average model to estimate the network delay after transmission. Using such information and its own information (i.e., local computing resource and energy availability), the user decides whether to offload its task to the MEC network via the BS. In case, a user decides to offload its task then the task deadline is not revealed to the MEC network to maintain fairness. Thereafter, the central controller of the MEC network performs optimal task allocation and notification of the computation results to the users. The impact of various user-parameters such as task generation probability, deadline, task size and processing density on the users and the MEC network are analyzed using extensive simulations.
{"title":"Decentralized Computation Offloading in Mobile Edge Computing Systems","authors":"Rohan Sharma, Kushaal Gummaraju, Pranav Anantharam, Ojaswi Saraf, Vamsi Krishna Tumuluru","doi":"10.1109/SmartNets50376.2021.9553007","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9553007","url":null,"abstract":"Existing works on mobile edge computing (MEC) which operate under multi-user and multi-server scenarios often assume centralized computation offloading. This paper proposes a decentralized computation offloading scheme where a user does not require information about other users and about the MEC network (e.g., number of servers, network topology). Under the proposed computation offloading scheme, a user infers the transmission delay from the link rate assigned by its associated base station (BS). Further, each user privately deploys a moving average model to estimate the network delay after transmission. Using such information and its own information (i.e., local computing resource and energy availability), the user decides whether to offload its task to the MEC network via the BS. In case, a user decides to offload its task then the task deadline is not revealed to the MEC network to maintain fairness. Thereafter, the central controller of the MEC network performs optimal task allocation and notification of the computation results to the users. The impact of various user-parameters such as task generation probability, deadline, task size and processing density on the users and the MEC network are analyzed using extensive simulations.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133625729","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}
In this paper, we propose a novel proactive content caching problem in which a set of contending edge service providers (ESPs) in a given region offer their storage and link capacities to the content provider (CP). The privacy of each contending ESP is preserved in the proposed caching problem. Each ESP independently determines the amount of storage and link capacity it can offer to the CP based on the local forecast of the content requests and its local edge resources. Unlike existing works, an ESP’s decision making problem is modeled as a robust mixed integer problem due to the uncertain storage capacity. Based on the offers made by the ESPs and its own prediction of the content requests, the CP determines the optimal content placement decisions while reserving the storage and link capacities under the ESPs in order to serve its clients at different bit rates. The CP also finds the optimal allocation of predicted requests across the ESPs. The decisions of the CP are found using a separate mixed integer problem which minimizes the payments given by the CP to the ESPs for their service. We show the impact of the robust parameter on the content placement decisions. We also perform sensitivity analysis of the CP’s decisions.
{"title":"Privacy-aware Robust Proactive Content Caching using Edge Service Providers","authors":"Rishi Kashyap, Manasa Bhat, Deepa Umashankar, Vamsi Krishna Tumuluru","doi":"10.1109/SmartNets50376.2021.9555417","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555417","url":null,"abstract":"In this paper, we propose a novel proactive content caching problem in which a set of contending edge service providers (ESPs) in a given region offer their storage and link capacities to the content provider (CP). The privacy of each contending ESP is preserved in the proposed caching problem. Each ESP independently determines the amount of storage and link capacity it can offer to the CP based on the local forecast of the content requests and its local edge resources. Unlike existing works, an ESP’s decision making problem is modeled as a robust mixed integer problem due to the uncertain storage capacity. Based on the offers made by the ESPs and its own prediction of the content requests, the CP determines the optimal content placement decisions while reserving the storage and link capacities under the ESPs in order to serve its clients at different bit rates. The CP also finds the optimal allocation of predicted requests across the ESPs. The decisions of the CP are found using a separate mixed integer problem which minimizes the payments given by the CP to the ESPs for their service. We show the impact of the robust parameter on the content placement decisions. We also perform sensitivity analysis of the CP’s decisions.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134344331","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 : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555424
K. Padmanabh, Ahmad Al-Rubaie, John Davies, Sandra Stincic, A. Aljasmi
The Chiller of a Heating, Ventilation, and Air-Conditioning (HVAC) system is a complex and expensive multicomponent appliance that is not impervious to failure. Predicting, or even identifying, a fault at its inception, can reduce the scale of the damage and mitigate the potential losses to be incurred, both financial and operational. This paper presents a systematic approach for the analysis of multiple streams of data from chillers to identify potential failures as soon as they become detectable from the data. The data streams are received from sensors in the IoT ecosystem of chillers to monitor the multitude of processes and parameters that are vital to their operation. Chillers have built-in mechanisms to generate alarms when key sensor values go beyond designated limits. A certain combination of these alarms is responsible for chiller failure, therefore, our proposed method needs to first predict these alarms using multi-sensor data fusion. Thus, in this IoT ecosystem there are two levels of sensor fusion for our predictive models: at the sensor level and at the derived alarms level. The final objective is to determine “time-time-to-next-alarm” TA). The model for TTA is built using time-shifted sensor values. Since chiller failure is a function of sensor alarms, and both are binary in nature, a special technique of logistic circuits is used to mimic the combination logical circuit to predict the failure of the chiller.
{"title":"Fault Prediction in HVAC Chillers by Analysis of Internal System Dynamics","authors":"K. Padmanabh, Ahmad Al-Rubaie, John Davies, Sandra Stincic, A. Aljasmi","doi":"10.1109/SmartNets50376.2021.9555424","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555424","url":null,"abstract":"The Chiller of a Heating, Ventilation, and Air-Conditioning (HVAC) system is a complex and expensive multicomponent appliance that is not impervious to failure. Predicting, or even identifying, a fault at its inception, can reduce the scale of the damage and mitigate the potential losses to be incurred, both financial and operational. This paper presents a systematic approach for the analysis of multiple streams of data from chillers to identify potential failures as soon as they become detectable from the data. The data streams are received from sensors in the IoT ecosystem of chillers to monitor the multitude of processes and parameters that are vital to their operation. Chillers have built-in mechanisms to generate alarms when key sensor values go beyond designated limits. A certain combination of these alarms is responsible for chiller failure, therefore, our proposed method needs to first predict these alarms using multi-sensor data fusion. Thus, in this IoT ecosystem there are two levels of sensor fusion for our predictive models: at the sensor level and at the derived alarms level. The final objective is to determine “time-time-to-next-alarm” TA). The model for TTA is built using time-shifted sensor values. Since chiller failure is a function of sensor alarms, and both are binary in nature, a special technique of logistic circuits is used to mimic the combination logical circuit to predict the failure of the chiller.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132347026","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 : 2021-09-22DOI: 10.1109/SmartNets50376.2021.9555422
Shruti Jadon, A. Patankar, Jan Kanty Milczek
Time-series forecasting has been an important research domain with significant applications, such as ECG predictions, sales forecasting, weather conditions, and recently COVID-19 spread predictions. Many researchers have investigated a multitude of modeling approaches to meet the requirements of these wide ranges of applications. In this context, our work focuses on reviewing different forecasting approaches for telemetry data collected in networks and data centers. Forecasting of telemetry data is a critical feature of network and data center management products. However, there are multiple options of forecasting approaches that range from a simple linear statistical model to high-capacity deep learning architectures. In this paper, we summarize and evaluate the performance of many well-known time series forecasting techniques. This research evaluation aims to provide a comprehensive summary for further innovation in forecasting approaches for telemetry data.
{"title":"Challenges and Approaches to Time-Series Forecasting for Traffic Prediction at Data Centers","authors":"Shruti Jadon, A. Patankar, Jan Kanty Milczek","doi":"10.1109/SmartNets50376.2021.9555422","DOIUrl":"https://doi.org/10.1109/SmartNets50376.2021.9555422","url":null,"abstract":"Time-series forecasting has been an important research domain with significant applications, such as ECG predictions, sales forecasting, weather conditions, and recently COVID-19 spread predictions. Many researchers have investigated a multitude of modeling approaches to meet the requirements of these wide ranges of applications. In this context, our work focuses on reviewing different forecasting approaches for telemetry data collected in networks and data centers. Forecasting of telemetry data is a critical feature of network and data center management products. However, there are multiple options of forecasting approaches that range from a simple linear statistical model to high-capacity deep learning architectures. In this paper, we summarize and evaluate the performance of many well-known time series forecasting techniques. This research evaluation aims to provide a comprehensive summary for further innovation in forecasting approaches for telemetry data.","PeriodicalId":443191,"journal":{"name":"2021 International Conference on Smart Applications, Communications and Networking (SmartNets)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132259670","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：