Deep Learning for the Internet of Things

Yao, Shuochao; Zhao, Yiran; Zhang, Aston; Hu, Shaohan; Shao, Huajie; Zhang, Chao; Lu, Shan; Abdelzaher, Tarek

doi:10.1109/mc.2018.2381131

Cited by 115 publications

(57 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their approach is based on unsupervised DL methods, in particular, the autoencoder architectures [90,91], where the encoding and decoding functions are parameterized by two CNNs and the communication channel is incorporated in the NN architecture as a non-trainable layer. Additionally, DL has recently applied to IoT systems [84][85][86][87][88][89], but none of them applies DL for solving the joint security-transmission problem.…”

Section: Overview Of Relevant Workmentioning

confidence: 99%

An innovative approach to integrate unequal protection-based steganography and progressive transmission of physiological data

2020

View full text Add to dashboard Cite

Evolving digital technologies in remote health monitoring require an energy-efficient method for secure and reliable transmission of patient's/user's confidential information from the sensor nodes to the cloud/medical server. Thus, a united scheme of the physiological signal steganography and its communication by benefitting from the unequal significance between different parts of the physiological data are emphasized. We believe higher steganography coding strength and more robust source-channel coding would protect extremely vital parts of the physiological data. Therefore, data integrity and transmission efficiency of packet information achieved in a resilient way. We formulate our idea of joint steganography-source-channel coding (JS 2 C 2) as an optimization problem to simultaneously securing and minimizing the transmission energy consumption. A low-complexity deep learning-based ECG classification algorithm along with its secure and energy-efficient neural JS 2 C 2 transmission for real-time monitoring has been realized. The optimal parameters for our united framework have been calculated by JS 2 C 2 optimization method. Our steganography algorithm unequal steganography embedding (USE) achieves very low wavelet-based weighted percent root-mean-squared difference lower than 0.5%. Furthermore, the high correlation between cover and stego and low end-to-end mean-square error (MSE) indicates resilient imperceptibility and maintains the diagnosability of the physiological signal. Moreover, low MSE between embedded and extracted data validates that embedded confidential data has been extracted with negligible distortion. In addition, for the given distortion, the USE-based framework's energy consumption is much smaller (by 55% in typical application scenario) as compared with the equal steganography embedding-based approach's energy consumption. Keywords Electrocardiography (ECG) • Unequal error protection (UEP) • Unequal steganography embedding (USE) • Joint steganography source coding (JS 2) • Joint source-channel coding (JSC 2) • Joint steganography-source-channel coding (JS 2 C 2) • Deep learning (DL) • Security • Correlation • End-to-end distortion • Energy efficiency

show abstract

Section: Overview Of Relevant Workmentioning

confidence: 99%

An innovative approach to integrate unequal protection-based steganography and progressive transmission of physiological data

2020

View full text Add to dashboard Cite

show abstract

“…Neglecting friction such as air resistance, all small bodies accelerate in a gravitational field at the same rate relative to the center of mass. [7][8][9] This is true regardless of the masses or compositions of the bodies. At different points on Earth, objects fall with an acceleration between 9.764 m/s 2 and 9.834 m/s 2 depending on altitude and latitude, with a conventional standard value of 9.80 m/s 2 (approximately 32.174 ft/s 2 ) [10].…”

Section: Introductionmentioning

confidence: 95%

Realistic Simulations of Non-Linear Acceleration of the Rocket in the Air

Xiong¹

2018

AJPA

View full text Add to dashboard Cite

The motion of the object in the medium has always been a hot research topic, and it is closely connected with many applications in our life. The acceleration of the object with multiple forces becomes very complicated, especially when these forces depend on the motion of the object. The exact formula for the object motion is a differential-integral equation and is very difficult to be solved analytically. One example of this kind of motions is the rocket launch. With sufficient thrust, the rocket can obtain an acceleration large enough to escape from the gravity of the earth. With the increasing height, the gravity from the earth becomes smaller, which affects the net acceleration of the rocket. Meanwhile, the air resistance becomes more and more important when the velocity of the rocket increases. It even plays the main role in the middle stage of the launch. Also, as the air resistance depends on both the velocity of the rocket and the air density (there is no air resistance in vacuum), the air resistance will decrease when the air density becomes small enough at the large height. In this article, a model that includes all of the factors mentioned above is established, and how these forces change the velocity of the rocket is analyzed. Two scenarios, one with air resistance and one without, are described. The velocity of the rocket in each scenario is represented by graphs, which are compared. With justification, the Taylor series is used to solve the differential-integral equation, and it is found that the fuel thrust and the gravity become important in the rocket launch at the beginning stage. In the middle stage, the air resistance begins to have a significant effect and reduces the acceleration of the rocket. In the final stage, there is virtually no gravity or air resistance, and only the fuel thrust contributes to the acceleration of the rocket.

show abstract

“…However, the adoption of DL in IoT-based architectures for AmI is currently limited due to high computational requirements and the need for large quantities of labeled training data. To adapt DL models to the computing and network limitations in IoT, recent research trends are considering methods to compress neural structures [42] or shift part of the computation to edge nodes [10]. To reduce the need for labeled data, recent methods are proposing the use of unsupervised DL techniques, that do not require labels, or Deep Reinforcement Learning, in which labels are assigned based on users' feedback [7].…”

Section: B Algorithmsmentioning

confidence: 99%

IoT-based Architectures for Sensing and Local Data Processing in Ambient Intelligence: Research and Industrial Trends

Cai

Genovese

Piuri

et al. 2019

2019 IEEE International Instrumentation and Measurement Technology Conference (I2MTC)

View full text Add to dashboard Cite

This paper presents an overview of new-generation technologies based on Internet of Things (IoT) and Ambient Intelligence (AmI), which create smart environments that respond intelligently to the presence of people, by collecting data from sensors, aggregating measurements, and extracting knowledge to support daily activities, perform proactive actions, and improve the quality of life. Recent advances in miniaturized instrumentation, general-purpose computing architectures, advanced communication networks, and non-intrusive measurement procedures are enabling the introduction of IoT and AmI technologies in a wider range of applications. To efficiently process the large quantities of data collected in recent AmI applications, many architectures use remote cloud computing, either for data storage or for faster computation. However, local data processing architectures are often preferred over cloud computing in the cases of privacy-compliant or time-critical applications. To highlight recent advances of AmI environments for these applications, in this paper we focus on the technologies, challenges, and research trends in new-generation IoT-based architectures requiring local data processing techniques, with specific attention to smart homes, intelligent vehicles, and healthcare.

show abstract

Deep Learning for the Internet of Things

Cited by 115 publications

References 56 publications

An innovative approach to integrate unequal protection-based steganography and progressive transmission of physiological data

An innovative approach to integrate unequal protection-based steganography and progressive transmission of physiological data

Realistic Simulations of Non-Linear Acceleration of the Rocket in the Air

IoT-based Architectures for Sensing and Local Data Processing in Ambient Intelligence: Research and Industrial Trends

Contact Info

Product

Resources

About