528
B. Farahani et al.
sharing, data purging, and data buffering. An edge node may also include, as
an example, signal processing, concurrent streaming, event handling, embedded
web server, embedded WebSocket server, etc.
• Streaming edge analytics: In some e-Health applications such as anomaly detection, it is a major necessity to learn actionable insight and actionable information
in real time close to the local context. Keeping this in mind, edge nodes should
be able to analyze the stream of device and sensor data with millisecond response
time. To do so, edge nodes can incorporate lightweight feature extraction,
data mining, time-series pattern recognition, machine learning, rule-based event
processing, and automated reasoning.
• Data delivery: Edge nodes can rely on either of the following message-exchange
techniques to deliver the IoT data: (i) message-based, (ii) request-based, and (iii)
publish-subscribe.
• Security and data protection: To protect patient data, the fog node offers multilayer security measures for authentication, encryption, and access control to fully
meet the requirements of FDA standards.
• Flexible integration: Considering the availability of different device vendors and
OEMs, edge nodes should implement a wide-range of interface standards to
maintain interoperability. To address the integration concern, edge nodes should
be compatible with a large variety of communication protocols and peripherals
(e.g., UART, SPI, and USB), PAN and WSN protocols (e.g., RFID, BLE, Zigbee,
Wi-Fi, 3G/4G, and Ethernet), and wired protocols (e.g., Ethernet).
• Protocol translation: Another challenge arises from the fact that there is a large
number of communication protocols at different levels of abstraction as follow:
– Network layer: An IoT e-Health network is scattered among various networking protocols (e.g., BLE, ZigBee, Wi-Fi). To bridge the gap among these
protocols, the edge node needs to convert and translate the incoming stream
to an appropriate format and propagate it to the destination network.
– Message layer: A large number of application-level protocols (e.g., MQTT,
CoAP, and XMPP) or processing messages exists. Thereby, it is very crucial
that edge nodes despite the underlying differences of standards be able to
transfer messages among different protocols.
– Data annotation layer: Different organizations proposed distinct standards for
integration, exchange, and retrieval of e-Health information (such as HL7).
Whenever it is required, edge nodes should be capable of understanding,
processing, and translating the data.
11.4.3 Cloud Layer
The cloud platform can benefit from a multi-layer architecture that consists of the
following layers (see Fig. 11.5) [1, 9].
B. Farahani et al.
sharing, data purging, and data buffering. An edge node may also include, as
an example, signal processing, concurrent streaming, event handling, embedded
web server, embedded WebSocket server, etc.
• Streaming edge analytics: In some e-Health applications such as anomaly detection, it is a major necessity to learn actionable insight and actionable information
in real time close to the local context. Keeping this in mind, edge nodes should
be able to analyze the stream of device and sensor data with millisecond response
time. To do so, edge nodes can incorporate lightweight feature extraction,
data mining, time-series pattern recognition, machine learning, rule-based event
processing, and automated reasoning.
• Data delivery: Edge nodes can rely on either of the following message-exchange
techniques to deliver the IoT data: (i) message-based, (ii) request-based, and (iii)
publish-subscribe.
• Security and data protection: To protect patient data, the fog node offers multilayer security measures for authentication, encryption, and access control to fully
meet the requirements of FDA standards.
• Flexible integration: Considering the availability of different device vendors and
OEMs, edge nodes should implement a wide-range of interface standards to
maintain interoperability. To address the integration concern, edge nodes should
be compatible with a large variety of communication protocols and peripherals
(e.g., UART, SPI, and USB), PAN and WSN protocols (e.g., RFID, BLE, Zigbee,
Wi-Fi, 3G/4G, and Ethernet), and wired protocols (e.g., Ethernet).
• Protocol translation: Another challenge arises from the fact that there is a large
number of communication protocols at different levels of abstraction as follow:
– Network layer: An IoT e-Health network is scattered among various networking protocols (e.g., BLE, ZigBee, Wi-Fi). To bridge the gap among these
protocols, the edge node needs to convert and translate the incoming stream
to an appropriate format and propagate it to the destination network.
– Message layer: A large number of application-level protocols (e.g., MQTT,
CoAP, and XMPP) or processing messages exists. Thereby, it is very crucial
that edge nodes despite the underlying differences of standards be able to
transfer messages among different protocols.
– Data annotation layer: Different organizations proposed distinct standards for
integration, exchange, and retrieval of e-Health information (such as HL7).
Whenever it is required, edge nodes should be capable of understanding,
processing, and translating the data.
11.4.3 Cloud Layer
The cloud platform can benefit from a multi-layer architecture that consists of the
following layers (see Fig. 11.5) [1, 9].