Intelligent IoT networks, their data and how AI makes them efficient: HSC Enterprise Computing Intelligent Systems
“Investigate the infrastructure requirements for an intelligent network in an enterprise with devices linked through an Internet of Things (IoT), including networks with servers, local and cloud storage, and end-point devices, and communication links that enable the efficient control and flow of data; explore collection, type, storage, processing, application and transmission of relevant and surplus data in an intelligent IoT network; explain how AI supports efficiency in an IoT network”
An intelligent IoT network needs end-point devices, gateways, servers, local and cloud storage, and communication links chosen for range, speed, power and reliability, plus strong security. Data is collected, typed, stored, processed, applied and transmitted, with surplus data filtered to save resources and reduce risk. AI supports efficiency through edge decisions, predictive maintenance, optimisation and anomaly detection.
Jump to a section
What this dot point is asking
You need to describe what an enterprise needs to run an intelligent network of IoT devices, explain the life cycle of the data (collection, type, storage, processing, application, transmission) including surplus data, and explain how AI makes the network efficient.
The answer
Infrastructure requirements
- End-point devices: sensors (temperature, motion, location, cameras), actuators (valves, locks, motors), smart meters and wearables. Often low-power with limited processing.
- Gateways and edge devices: collect data from nearby end points, translate protocols, filter and aggregate data, and run local processing.
- Networks with servers: on-site or cloud servers that run applications, dashboards, databases and AI models.
- Local and cloud storage: local storage buffers data and supports fast or offline operation; cloud storage scales for long-term data and analytics.
- Communication links: short range (Bluetooth Low Energy, Zigbee, Wi-Fi), long range low power (LoRaWAN), cellular (4G, 5G), and wired (Ethernet, fibre). Chosen for range, bandwidth, latency, power use and reliability to enable efficient control and flow of data.
- Security: device authentication, encryption, network segmentation and firmware updates, because IoT devices are common attack targets.
The data in an IoT network
- Collection: continuous streams, periodic readings, or event-triggered readings.
- Type: numeric sensor values, location, images and video, audio, status flags. Mostly time-stamped, structured or semi-structured.
- Storage: hot storage for recent data, archives for history, with retention policies.
- Processing: at the edge (filtering, aggregation, real-time decisions) and in the cloud (analytics, machine learning).
- Application: automatic control, alerts, dashboards, predictions.
- Transmission: only what is needed, compressed and encrypted.
Relevant data serves the system's purpose. Surplus data is extra (duplicate readings, excess detail, data about people not needed for the task). It costs bandwidth, storage and energy, can slow analysis, and increases privacy and security risk, so good design filters it early and deletes it on schedule. Some surplus data may later prove useful for new analysis, which is why enterprises sometimes keep samples.
How AI supports efficiency
- Edge AI decides locally in milliseconds (a camera detects a hazard without sending video to the cloud).
- Predictive maintenance forecasts failures from sensor patterns, reducing downtime.
- Optimisation adjusts systems for energy, routing or scheduling (smart building heating that learns occupancy).
- Anomaly detection spots faults or cyberattacks in data streams.
- Data reduction uses AI to decide which data is worth sending or keeping.
A council installs smart bins that report how full they are.
- End points: ultrasonic fill-level sensors in 2,000 bins, battery powered.
- Links: LoRaWAN, which has long range and very low power use; sensors send a small reading every hour.
- Storage and servers: readings go to a cloud platform with a dashboard for the waste team.
- Surplus data: hourly readings of an unchanged bin are surplus; the sensor sends only when the level changes by 10% or more.
- AI: a model predicts when each bin will be full and plans the most efficient collection routes daily, cutting truck kilometres.
- Sending all data to the cloud by default
- Edge processing reduces delay, bandwidth and privacy risk.
- Ignoring communication link choice
- Justify links by range, bandwidth, latency and power.
- Treating surplus data as harmless
- It has costs and privacy risks.
Practice questions
Original practice questions graded from foundation to exam level, each with a full worked solution. Try them before revealing the solution.
foundation3 marksIdentify the infrastructure needed for an IoT network that monitors refrigerator temperatures in 50 supermarket stores.Show worked solution →
- End-point devices: temperature sensors in every fridge and freezer.
- Communication links: wireless (Wi-Fi or Zigbee) from sensors to an in-store gateway, then broadband or 4G to the cloud.
- Servers and storage: a gateway with local storage to buffer readings if the internet drops, and cloud servers and storage for analysis, dashboards and alerts.
Marking guide: 1 mark each for devices, communication links and servers/storage.
core4 marksSensors in a smart building record temperature every second. Explain how the enterprise should manage relevant and surplus data from collection to transmission.Show worked solution →
- Collection
- reading every second produces mostly identical values (surplus data). The sensor could record every minute or only when the temperature changes by more than 0.5 degrees.
- Processing at the edge
- the gateway averages readings and forwards only summaries and exceptions, reducing transmission.
- Storage
- recent detailed data is kept locally for a short time; long-term storage in the cloud keeps hourly averages, with a retention policy that deletes raw surplus data.
- Application
- relevant data (trends, exceptions) drives heating and cooling decisions and energy reports.
Marking guide: 1 mark each for collection, processing, storage and application or transmission decisions.
exam6 marksA mining company operates driverless haul trucks and thousands of sensors on remote sites. Explain the infrastructure requirements for its intelligent IoT network and how AI supports efficiency.Show worked solution →
- Infrastructure
- End-point devices include truck sensors (GPS, engine temperature, tyre pressure, cameras, lidar) and fixed sensors on conveyors and crushers. A private LTE or 5G network provides low-latency, reliable links across the site, because driverless trucks need real-time control. On-site edge servers process time-critical data (collision avoidance, traffic control) and store data locally in case satellite or fibre links to head office fail. Cloud storage and servers at head office hold historical data for fleet-wide analysis. Security (encryption, device authentication, network segmentation) protects control systems.
- AI and efficiency
- On the edge, AI models detect obstacles and plan routes in real time. Predictive maintenance models learn from vibration and temperature patterns to schedule repairs before breakdowns, reducing downtime. Optimisation algorithms dispatch trucks to minimise waiting at loaders and fuel use. Anomaly detection flags unusual sensor readings that may indicate faults or security threats.
- Overall
- The network must be fast, reliable and secure at the edge while feeding the cloud for long-term learning; AI turns the data into safer, cheaper and more productive operations.
Marking guide: 3 marks for infrastructure (devices, links, servers and storage), 3 marks for AI efficiency examples.