Font size
WorksheetsEdge Quiz
Total questions: 158
Worksheet time: 1hrs 23mins
What is Cloud Computing?
It is a way to access software and services over the internet.
It is a way to physically connect multiple computers together
It is a way to protect data from unauthorized access.
It is a way to store data on your local computer.
Which of the following is NOT an advantage of Edge Computing?
Low investment cost
Significantly reduced data storage requirements
Reduces Network Bandwidth Requirements
Faster response time
Based on the lecture, which of the following is not a typical deployment challenge for Edge computing devices?
Power
Waterproofing
Thermal
Sound proofing
Which of the following is a good example of an AIoT application?
A smart home thermostat that learns your preferences over time
A fitness tracker that monitors your heart rate and sleep patterns
A wearable device that tracks your steps and calories burned
A wireless security camera that sends alerts to your smartphone
What is the purpose of AIoT?
To improve the design of IoT devices
To integrate artificial intelligence with internet of things
To replace IoT devices with AI-powered devices
To create a secure network for IoT devices
Match the Edge computing platforms to how they perform inferencing on the device.
GPU-centric
CPU-centric
Raspberry Pi
NVidia Jetson
Match the Edge computing platforms to how they perform inferencing on the device.
CPU-centric
NVidia Jetson
GPU-centric
Raspberry Pi
Which of the following is not a benefit of AIoT? (pick the best answer)
Enhanced security
Improved energy efficiency
Increased productivity
Decreased complexity
AIoT stands for?
Artificial Internet of Things
Advanced IoT
Artificial Intelligence of Things
Artificial Intelligence on IoT
Which of the following is a limitation of AIoT?
It requires a high degree of technical expertise to implement
It can be expensive to implement
It is only suitable for large-scale applications
It is limited by the processing power of IoT devices
Which of the following is TRUE about Fog Computing?
Fog provides low latency
Fog computing are located nearer to the Cloud computing
Fog computing is a separate architecture that can replace cloud computing.
Cloud is a centralized system, while the fog is a distributed decentralized infrastructure
In BLE terminology, what is a 'GATT'?
Generic Audio Transmission Techno
Generic Attribute Profile
Generic Authentication Transfer Tool
Generic Access Transmission Technique
LoRa's spreading factor affects which of the following?
Power consumption only
Operating frequency
Data rate and range
Modulation technique
Which standard does HaLow follow for wireless communication?
802.11ah
802.11b
802.11ac
802.11n
Which Sub-GHz ISM frequency does LoRa operate in Singapore?2.4 Ghz
2.4 Ghz
868 Mhz
466 Mhz
923 Mhz
Link Budget …
refers to the amount of loss that a data link (transmitter to receiver) can tolerate in order to operate properly
refers to the amount of loss that a data link (transmitter to receiver) should have.
refers to the difference between the minimum expected power received at the receiver's end, and the receiver's sensitivity.
refers to a connection between two entities
In LoRa, which class does the following end device fall under: Battery-powered actuators, Energy efficient with latency controlled downlink, Slotted communication synchronized with a beacon.
D
B
C
A
. Which of the following do not contribute to routing in a ZigBee network?
ZigBee Router
ZigBee End Device
ZigBee Master
ZigBee Coordinator
Bluetooth is designed for short broadcasts. There are forty 2-Mhz channels. How many of these channels are reserved for advertising? (enter numerical number)
(a)
What is the maximum data rate of the latest BLE?
1 Mbps
3 Mbps
2 Mbps
4 Mbps
What is the role of Zigbee end devices in the Zigbee network?
Communicates with other nodes
Controls the network
Both
None
Z-Wave is an implementation of a ZigBee protocol.
True
False
Which frequency does Z-Wave operate on?
433 Mhz
868 Mhz
2.4 Ghz
5 Ghz
Which class of LoRaWAN is designed for bidirectional end-devices with scheduled receive slots?
Class C
Class D
Class A
Class B
What is the purpose of Zigbee communication?
Healthcare
Industrial Control
Home Automation
All of the above
In LoRa, what does increasing the spreading factor (SF) do to the signal's range and data rate?
Increases range and decreases data rate
Decreases both range and data rate
Decreases range and increases data rate
Increases range and data rate
Which of the following is not a feature of Zigbee? (choose the best one)
High Security
Low Power Consumption
Real-time Communication
Large Bandwidth
What is a characteristic of BLE's 'Advertising' mode?
Secure data encryption.
b. High data transfer rate.
Device discovery and connection establishment.
Long-range communication.
What does 'Pairing' in BLE involve?
What does 'Pairing' in BLE involve?
Exchanging security keys for secure communication.
Connecting two devices for data transfer.
Adjusting the power level for optimal connection.
What type of modulation is used in LoRa?
Amplitude Modulation (AM)
Chirp Spread Spectrum (CSS)
Phase-Shift Keying (PSK)
Frequency Shift Keying (FSK)
What is the range typically achievable with BLE technology?
Up to 50 meters
Up to 150 meters
Up to 100 meters
Up to 10 meters
LoRa modulation is a variation of ...
Chord Spread Spectrum
Chord Space Spectrum
Chirp Spread Spectrum
Chirp Spectrum Sector
LoRa modulation technique sits at which OSI layer?
Physical
Transport
Presentation
Session
Application
Which of the following supports low energy radio operation?
HaLow
Bluetooth
BLE
WiFi
Which Class of LoRaWAN is best suited for applications requiring minimal latency?
Class B
Class A
Class D
Class C
What does LoRaWAN stand for?
Long Range Wide Area Network
Logical Routing WAN
Low Range Wide Area Network
Low Radiation Wireless Access Network
What is the operating frequency of HaLow?
5 Ghz
900 Hz
2.4 Ghz
Sub-1 Ghz
What is the main function of the Device EUI in LoRaWAN?
To define the device's operating frequency
To specify the data rate
To encrypt data
To identify the device uniquely on the network
LoRaWAN's network server is responsible for:
Processing and forwarding data to application servers
Managing the radio frequency spectrum
Controlling the power output of end devices
Directly interacting with end-user applications
Which network topology is not supported by ZigBee?
Bus
Star
Cluster Tree
Mesh
Which of the following is needed to be a "ZigBee Compliant Product"
ZigBee Stack
IEEE 802.15.4
IEEE 802.11
Application Profile
Zigbee have a transport layer.
True
False
What is the main advantages of using LoRa for IoT applications?
High Data Rate
Low Data Rate
Low Power Consumption
Long Range
In BLE, what is the role of a 'Central' device?
It only receives data
It connects to and controls peripheral devices.
It acts as an intermediary between two devices.
It broadcasts data to peripheral devices.
Which of the following Topology does Bluetooth 4.0 support?
Mesh
Scatter Net
Star
Ring
What does LoRa stand for?
Long Resonance
Low Range
Long Radio
Long Range
What is the typical range of HaLow compared to traditional Wi-Fi?
Same range
Shorter range
Longer range
Unlimited range
Which frequency band does BLE operate in?
5.0 GHz
2.4 GHz
433 MHz
868 MHz
915 MHz
LoRaWAN modulation technique sits at which OSI layer?
Application
Physical
Network
Session
Datalink
LoRa is primarily used for?
Long-range, low-power consumption
Short-range communication
Audio and video transmission
Faster speeds
Which feature distinguishes BLE from Classic Bluetooth?
Longer range
Lower energy consumption.
Higher data transfer rates.
Better audio quality.
BLE supports which kind of communication topology?
Point-to-Point
Star
Bus
Mesh
Which of the following schemes is used by Bluetooth?
Frequency hopping FDD scheme
DSSS FDD scheme
Frequency hopping TDD scheme
DSSS TDD scheme
Pyramid scheme
What does BLE stand for?
Basic Link Encryption
Bluetooth Long Engagement
Bluetooth Low Energy
Bluetooth Low Efficiency
What is the role of Zigbee routers in the Zigbee network?
Acts as a relay between other nodes
Controls the network
Both
None
BLE is particularly designed for:
Audio streaming
Long-distance communication
High-speed data transfer
Short-range, low-power communication
What is the main drawback of using LoRa for IoT applications?
High power consumption
Limited data rate
Long range
Short range
What is the role of Zigbee coordinator in the Zigbee network?
Controls the network
Communicates with other nodes
Both
None
What is the HTTP method used to create a new resource in RESTful architecture?
PUT
DELETE
POST
GET
What is the default port number for MQTT?
8080
1833
80
8883
What is a topic in MQTT?
A string used to filter messages in a publish/subscribe model
A unique identifier for a message
A unique identifier for a message
A message payload
MQTT is _______ oriented.
Data
Device
Message
Network
MQTT transmits data to and from a broker in which format. (Choose the best option)
Text based data
XML
JSON
Binary data
What type of protocol is MQTT?
Application protocol
Network protocol
Transport protocol
Presentation protocol
What is a resource in the context of RESTful web services?
An entity that can be accessed and manipulated through a URI
A function that performs a specific action
A function that performs a specific action
A set of procedures for accessing a database
In RESTful APIs, what is used to uniquely identify a resource?
API key
URI (Uniform Resource Identifier)
Session ID
Query parameters
In RESTful APIs, what does a status code in the 5xx range signify?
Server error
Redirection
Successful operation
Client error
Which protocol is CoAP closely modeled after?
HTTP
AMQP
FTP
MQTT
CoAP Protocol is comprised of which sub-layers?
UDP
Request & Response
Observation
Message
What is the maximum number of clients that can subscribe to a single topic in MQTT?
10
1
100
Unlimited
What does CoAP stand for?
Compact Application Protocol
Constrained Application Protocol
Common Application Protocol
Control Application Protocol
What does MQTT stand for?
Message Querying Telecommunication Transmission
Message Queuing Transaction Transport
Message Queuing Telemetry Transport
Message Quorum Transport Technology
Which layer is CoAP?
Transport layer
Application layer
Service layer
Control layer
CoAP is specialized in?
Wired applications
Internet applications
Wireless applications
Device applications
What transport layer protocol does CoAP use?
ICMP
REST
UDP
TCP
CoAP supports which of the following methods similar to HTTP?
CONNECT, TRACE, PATCH
None of them
OPTIONS, HEAD
GET, POST, PUT, DELETE
CoAP messages can be of which types?
Confirmable and Non-Confirmable
Request and Response
Synchronous and Asynchronous
Persistent and Transient
CoAP is designed for which kind of networks?
Resource-constrained networks
Datacenter Networks
Wide Area Networks (WAN)
Local Area Networks (LAN)
Which of the following is NOT a constraint of RESTful architecture?
Cookie-based authentication
Cacheability
Statelessness
Statelessness
What does idempotence mean in the context of RESTful APIs?
The API produces different results with each reques
The API can handle asynchronous requests
The API encrypts data for security
The API produces the same result no matter how many times the same request is made
The core of the protocol is specified in ______?
RFC 7452
RFC 7252
RFC 7524
RFC 7254
To receive a retained message?
You must keep the message.
You must subscribe with the retained message set
You must subscribe to a topic that has been published with the retained message set.
You must subscribe with a QOS of 1 or 2
Which HTTP status code indicates a successful GET request in a RESTful API?
400 Bad Request
201 Created
200 OK
500 Internal Server Error
CoAP is a specialised _________ protocol.
Power
Web transfer
Resource
Application
Which of the following QoS levels in MQTT ensures message delivery at least once?
QoS 0
None of them
QoS 2
QoS 1
Which of the following is not a characteristic of MQTT?
Quality of Service (QoS)
Request/response messaging model
Low overhead
Publish/subscribe messaging model
he Last Will Message is used to:
Gain inheritance
Notify a publisher of a network failure
Guarantee that the message was delivered
Notify clients of a failure in the publisher
Which of the following is not a client type in MQTT?
Broker
Publisher
Producer
Subscriber
In RESTful APIs, what is the purpose of a payload in a POST request?
To provide additional headers
To deliver the client's request to the server
To contain the data being sent by the client
To carry control information
What is the primary role of a RESTful API's Uniform Interface constraint?
To define specific operations like GET and POST
To separate client and server implementations
To ensure that the API is stateful
To standardize the way clients interact with the server
Which HTTP method should be used to update an existing resource in RESTful web services?
PUT
GET
POST
DELETE
MQTT uses the publish/subscribe model. How does the publisher know who is subscribing?
The subscriber informs the publisher
Publisher uses the "Eye of Sauron"
The publisher doesn't know
The publisher gets notified when they subscribe.
CoAP provides which of the following requirements?
Simplicity and low overhead
Lower overhead and multicast support
Multicast support and simplicity
Multicast support, lower overhead and simplicity
What mechanism does RESTful APIs use to indicate the current state of the resource?
HTTP headers
Response body
HTTP status codes
API tokens
Select the TRUE statement(s) for CoAP protocol?
Uses GET message requests to retrieve subscription
Uses FETCH message requests to retrieve subscription.
Uses PUSH message requests to create subscription.
Uses PUT message requests to create subscription.
What does REST stand for?
Real-time State Transfer
Representational State Transfer
Remote Server Transaction
Relative State Transmission
Which of the following HTTP methods is used to retrieve a resource in RESTful architecture?
PUT
GET
POST
DELETE
What type of architecture does RESTful APIs follow?
Peer-to-peer architecture
Monolithic architecture
Client-server architecture
Publisher-Subscriber architecture
Microservices architecture
Which of the following is not a valid time domain feature?
Root Mean Square Energy
Zero Crossing Rate
Spectral Centroid
Amplitude Envelope
What is the typical audio sampling rate?
44.1 kHz
20 kHz
88 kHz
22 kHz
Which of the following is an important feature for speech recognition?
All of the above
Edges
Histogram of gradients
Spectral features
Which type of input will be best suited as an input to a machine learning algorithm?
Raw audio signal
Spectrogram of audio signal
Spectrum of audio signal
Filtered audio signal
What is the importance of Mel-scale?
Sounds of equal distance on the Mel Scale are perceived to be of equal distance to humans
It is easier for a machine learning algorithm to tackle
It is easier for a machine learning algorithm to tackle
It is a form of compression of audio signals
Which of the following sound frequencies will a normal human able to distinguish better?
10000 to 10200 Hz
1000 to 1200 Hz
5000 to 5200 Hz
100 to 200 Hz
What is the need of a power spectrogram?
What is the need of a power spectrogram?
None of the above
To capture the change of frequency of a signal with time
To capture the change of amplitude of a signal with time
What is the typical fundamental frequency range of adult voice?
31 to 19000 Hz
Below 85 Hz
85 to 155 Hz
165 to 255 Hz
What is Cepstrum?
It is a tool to find anomalies in the signal
All of the above
It is a tool to find frequencies in the signal
It is a tool for finding spectrum of a spectrum
Why YAMNet is a suitable sound classification model for edge applications?
It is a decision tree based approach
It uses multiple sound features as input
None of the above
It is based on light weight MobileNet v1
Speech signals can be considered as which of the following type?
Stationary signals
Non-stationary signals
Random signals
None of the above
Why audio analytics on the edge is an important task?
Sensitivity of the data is critical for many applications
Redundancy of the data can be reduced (enable compression)
ast action (e.g. audio scene classification)
All of the above
What does a spectrum of a signal provide?
A measure of noise in the signal
All of the above
A measure of frequency components present in the signal
A measure of amplitude of the signal
Which filter type attenuates frequencies higher than the cutoff point?
Low-Pass
Notch
High-Pass
Band-Pass
Why are Mel-frequency cepstral coefficients (MFCC) commonly employed in speech processing?
All of the above
It emphasize features of the audio signal that are important for human speech
It reduces the redundancy in speech processing
It is one of the most efficient representations of speech for employing as an input to machine learning
What will be the total size of a HoG feature vector if a 9 bin approach based on 8x8 window which is normalized for 16x16 block?
256 x 1
36 x 1
9 x 1
8 x 1
Given an extremely resource constrained edge device (e.g. Raspberry Pi Pico) which is the following can be employed for hand land mark detection?
A convolutional neural network (CNN) model
A fully connected neural network mode
Image features + support vector machine (SVM)
Any of the above
In a gray scale image which is uint8 type, what will be the value corresponding to gray color?
0
255
124
None of the above
What does image gradient infer?
Low frequency in the image
Smooth texture in the image
Colors in the image
All sudden transitions (edges) in the image
From the matrix size below, identify a color image
256 x 256 x 3
256 x 256
256 x 256 x 2
256 x 245 x 5
Which of the following is useful for extracting depth from a video?
Local binary descriptor
Edges
Optical flow estimations
All of the above
Which of the following is the definition of histogram of an image?
A measure of frequency of occurrence of image intensities
A measure of frequency of occurrence of edges in the image
A measure of noise in the image
All of the above
Select a suitable window size for obtaining large image gradients in the image of size 256 x 256
3 x 3
8 x 8
64 x 64
256 x 256
Which is supposed to be a high frequency component in an image?
Random noise
Edges in image
Both a and b
None of the above
What is the preferred algorithm for hand gesture based interface?
Direct hand landmark detection – gesture control
Hand detection – Hand landmark detection – gesture control
Direct gesture control
None of the above
What is the starting point of any edge detection?
Smoothing followed by gradient operation
Gradient operation alone
Gradient followed by smoothing operation
All of the above
Which of the following feature(s) can be used for face detection in the image on an edge device?
Histogram of oriented gradients
Histogram of gradients
Histogram of oriented textures
None of the above
Which of the following images has the highest resolution?
256 x 256
2048 x 2048
1024 x 1024
16 x 16
Which is preferable edge only image analytics?
Image segmentation
Fingerprint informatics
Object detection
All of the above
Which is the following is a reason for increased popularity for computer vision on the edge?
Low latency is critical for fast action (e.g. motion segmentation)
Sensitivity of the data is critical for many applications
Redundancy of the data can be reduced (video summarization)
All of the above
Which of the following condition is apt for Pruning?
The histogram of the weights centered around zeros
A bi-modal Gaussian distributed histogram
A Poisson distributed histogram
None of the above
Which of the following is a representation of linear symmetric quantization? (x is input, x_int is output, s is scale, z is zero point, b is number of bits
x_int = clamp(round(x/s) + z , 0, 2^b – 1)
x_int = clamp(round(x/s) + z , -2^b -1, 2^b – 1)
x_int = clamp(round(x/s) , 0, 2^b – 1)
None of the above
What is a neural network distillation?
The learned teacher model is Pruned to form a student model
The learned teacher model is Quantized to form a student model
Learning a student model with low complexity from a teacher model
None of the above
Pruning of neural networks can be done in which of the following ways
Removing some of the weights
Reducing the bit-resolution of the weights
Reducing the spatial resolution of the input
All the possibilities
Which of the following cases is not useful for pruning of neural networks?
A very deep neural network
Pruning works in all cases
A recursive neural network
A modular network with the output of one neural network is input to the next
How to make rounding error smaller?
Decrease q_max
Increase q_max
Decrease q_min
All of the above
How do we find the scale factor and zero point of inputs/activations?
Not possible
Run representative mini-batches
Derive from theory
Predict from the trained model
What will be the scale factor for a data with minimum value of -50 and maximum value of 150 and for 8-bit resolution.
200
0.7843
0.0039
None of the above
What will be the dequantization function for a non-linear quantization (e.g. weight sharing)
x' = s (x_int – z)
x' = s*x_int
Needs a look up table (or code book)
Not available
What will be the starting point for weight sharing by combining Pruning with Quantization?
k-Means based clustering of weights
Huffman Coding of Weights
Quantization of Weights
Pruning of Weights
What will be the typical shape of a histogram after Pruning process?
A Poisson distributed histogram
A normal distributed histogram
A bi-modal Gaussian distributed histogram
Any of the above
Which of the following has the highest complexity of implementation?
Symmetric Signed Quantization
Symmetric Unsigned Quantization
Asymmetric Unsigned Quantization
All the above are equal in complexity
What will be the effective bit size for a quantization with level set {-1, 0, 1}?
3 bits
1.58 bits
1 bit
2 bits
Which of the following is useful for implementing a deep neural network on the edge device?
Higher number of parameters
Neural network weights represented by 32 bit fixed point representation
None of the answers
Both the higher parameters and 32 bit fixed point representation
Which of the below cases provides the best results?
L2 regularization without retrain
L1 regularization without retrain
L1 regularization with retrain
L2 regularization with iterative retraining
There are a plenty of benchmark datasets for deep learning which are targeted at edge devices
True
False
The aggregation of dataset and models in an Edge AI environment is known as Edge (a)
Which of the following languages/software is preferred for implementing on an edge device?
Python
Any of the following
C or C++
Matlab
Which of the following software development kit is dedicated to STM32 microcontrollers?
Nvidia TensorRT
AMD DNNDK
None of the following
X-Cube AI
Which of the following is a software tool for edge DL inference?
ExecuTorch
All of them
Tensorflowlite
Onnx
Which of the following conditions is not important for an Edge AI deployment?
Simultaneous design of algorithms with hardware
All of the following
The environmental conditions of edge AI deployment doesn’t really matter
Tools for automatic mapping of trained models
What is Edge Fleet Management?
None of the following
It is a hub where all the edge devices will publish/subscribe the data
It is a network of edge devices
It is a process of compressing DL algorithms to be run on an Edge Device
Which software tool supports the Raspberry Pi for edge inference?
Embedded Learning Library (ELL)
Intel OpenVINO Toolkit
None of the following
Matlab Deep Learning HDL Toolbox
The process of identifying the memory requirements and latency for running an ML algorithm on a specific hardware is known as?
(a)
It is not feasible to implement an unsupervised ML algorithm like k-Nearest Neighborhood (k-NN) on an edge device.
True
False
What is the trimmed version of Python 3.x which could be run on an edge device?
(a)
It is possible to run an ML algorithm on a field programmable gate device (FPGA) or an application specific integrated circuit (ASIC)
True
False
There are plenty of software frameworks available for neural architecture searches for edge devices.
True
False
In Hierarchical Federated Learning, which of the following is correct?
The model is trained at the edge
The model is trained in the cloud
The model is trained in the cloud and at the edge
None of the above
When the dataset is not shared, but the model is locally trained and shared across multiple compute devices, the framework is known as
Localized Learning
Federated Learning
Globalized Learning
Edge Learning
