Analytics & ML · Study deck

Edge Acquisition: Power Budgets and Pipelines

An edge pipeline processes sensor data near the device.

Data Dora is your guide for this deck.

edge
Data Dora, the module guide, in a scene from this chapter.
iotclass.org

After studying this chapter

Learning objectives

You will be able to:

  • Build a power budget for an edge acquisition node.
  • Design an IMU aggregation pipeline from energy and latency limits.
  • Explain: An edge pipeline processes sensor data near the device.
  • Explain: Mains power permits continuous acquisition while edge work primarily controls bandwidth; a battery branch makes wake time and radio duty cycle explicit; harvesting adds a variable energy budget that must adapt sampling and transmission.
iotclass.org

Major section

Start With the Decision

An edge pipeline processes sensor data near the device.

  • A window is the slice of readings it handles at once.
  • A model is the rule or learned pattern applied to those readings.
  • A fast pipeline is useless if its node runs out of power before service day.
Massive IoT and critical IoT use the same device, connectivity, and acquisition layers with opposing scale and latency priorities.
Massive IoT and critical IoT use the same device, connectivity, and acquisition layers with opposing scale and latency priorities.
iotclass.org

Major section

Power Budget Decision Tree

Mains power permits continuous acquisition while edge work primarily controls bandwidth; a battery branch makes wake time and radio duty cycle explicit; harvesting adds a variable energy budget that must adapt sampling and transmission.

  • The branches show why the same sensor workload needs different scheduling evidence in different deployments.
Power budget decision tree: mains-powered devices optimize for bandwidth, battery-powered devices duty-cycle sampling and transmission, and energy-harvesting devices adapt to available energy.
Power budget decision tree: mains-powered devices optimize for bandwidth, battery-powered devices duty-cycle sampling and transmission, and energy-harvesting devices adapt to available energy.
iotclass.org

Major section

Edge Processing Savings Calculator

Adjust the parameters below to see how edge processing affects connectivity costs for a multi-shed IoT deployment.

  • The dominant cost driver is typically the highest-bandwidth device (cameras).
  • Under this illustrative dairy-shed scenario’s stated prices, the modeled 99% data reduction makes rural 4G connectivity economically viable.
  • A single edge gateway serves all three roles, and the dominant cost driver is the highest-bandwidth device category (cameras, in this case).
iotclass.org

Major section

Edge Acquisition Storage Tiers

Edge acquisition design does not stop at the gateway.

  • The storage plan should be tied to a TCO calculation.
  • If the edge system only shifts cloud cost into unmanaged local maintenance, the architecture is not actually cheaper.
  • Retention policies should be executable, not just documented.
iotclass.org

Major section

Concept Relationships

Data generation patterns vary 1000x across categories (door sensors: bytes/day vs cameras: gigabytes/day).

  • Edge Compute Patterns: Where to process (edge/fog/cloud) depends on device capabilities established here.
  • Edge Fog Computing: Big Things can participate in fog layer; Small/Non-IP Things need edge gateways.
  • Data Quality and Preprocessing: Edge acquisition lets validation catch errors near the source before they propagate into cloud processing.

Why it matters

Edge Data Acquisition: Sampling and Compression: How to reduce data volume based on device category (cameras need compression, temperature sensors need aggregation).

iotclass.org

Deck summary

Key takeaways

An edge pipeline processes sensor data near the device.

  • Mains power permits continuous acquisition while edge work primarily controls bandwidth; a battery branch makes wake time and radio duty cycle explicit; harvesting adds a variable energy budget that must adapt sampling and transmission.
  • Adjust the parameters below to see how edge processing affects connectivity costs for a multi-shed IoT deployment.
  • Edge acquisition design does not stop at the gateway.
  • Data generation patterns vary 1000x across categories (door sensors: bytes/day vs cameras: gigabytes/day).
iotclass.org

Retrieval practice

Recall check 1 of 3

Data Dora says: answer from memory, then check your reasoning.

Q1Complete the IMU edge aggregation for data reduction:

Aself.buffer = []
Bself.buffer = {}
Cself.buffer = None
Dself.buffer = 0
Show answer

Answer: A The magnitude of a 3D acceleration vector is sqrt(x^2 + y^2 + z^2), giving a single scalar representing total acceleration regardless of orientation.

iotclass.org

Retrieval practice

Recall check 2 of 3

Data Dora says: answer from memory, then check your reasoning.

Q2A factory has temperature sensors using Modbus RS-485 protocol, IP cameras with Wi-Fi, and a central SCADA server. How would you classify these three device types?

ABig Things for all three because the deployment is in an industrial site.
BClassify sensors as Small IP, cameras as Big Thing, and SCADA as Non-IP because data size determines each interface.
CNon-IP Things for all three because each is used in industry.
DNon-IP temperature sensors, Small IP cameras, and Big Thing SCADA.
Show answer

Answer: D see answers page

iotclass.org

Retrieval practice

Recall check 3 of 3

Data Dora says: answer from memory, then check your reasoning.

Q3Why do Non-IP Things typically generate the least data volume despite often being the most numerous devices in IoT deployments?

AThey have slower processors that cannot generate data quickly
BGateways compress their data before counting it
CNetwork protocols limit how much they can transmit
DThey typically measure simple physical parameters
Show answer

Answer: D The chapter's simple temperature and door-state examples create small or sparse source records, while video frames are much richer.

Q4What aggregates fleet history for longer horizon analysis?

ACloud backend
BSensor and actuator layer
CEdge processing node
Show answer

Answer: A The cloud backend aggregates fleet history and coordinates longer-horizon analysis.

iotclass.org

Print reference

Answers 1 of 3

Answer key.

  1. A · The magnitude of a 3D acceleration vector is sqrt(x^2 + y^2 + z^2), giving a single scalar representing total acceleration regardless of orientation.
iotclass.org

Print reference

Answers 2 of 3

Answer key.

  1. D · The classification depends on connectivity capabilities, not the deployment environment: - Modbus RS-485 temperature sensors = Non-IP Things (no native IP stack, need gateway) - Wi-Fi IP cameras = Small IP Things (embedded device with IP connectivity) - SCADA server = Big Thing (full computer/database system) This classification determines the acquisition architecture: sensors need a gateway for protocol translation, cameras connect directly via Wi-Fi, and the SCADA server can use any IP protocol.
iotclass.org

Print reference

Answers 3 of 3

Answer key.

  1. D · The chapter's simple temperature and door-state examples create small or sparse source records, while video frames are much richer.
  2. A · The cloud backend aggregates fleet history and coordinates longer-horizon analysis.
iotclass.org