The Internet of Things (IoT) is a network of physical devices embedded with sensors, processors, and wireless connectivity that collect and exchange data over the internet without direct human interaction. IoT Analytics counted 15.9 billion connected devices globally in 2024, with projections of 32 billion by 2030. This guide covers IoT architecture, communication protocols, consumer and industrial applications, and the 3 primary security risks with documented examples.
What Is the Internet of Things?
The Internet of Things is a system in which physical objects — sensors, appliances, vehicles, machines — connect to IP networks and communicate data without requiring manual input. The term was coined by Kevin Ashton in 1999 at MIT’s Auto-ID Center in the context of RFID supply chain tracking.
The core requirement is that a device must have: a unique identifier (IP address or device ID), at least one sensor or actuator, and network connectivity. IoT devices range from a $5 temperature sensor transmitting over LoRaWAN to a $50,000 CNC machine streaming vibration data for predictive maintenance.
IoT Architecture: 4 Layers
IoT systems are organized into 4 functional layers, each handling a distinct aspect of data flow from physical sensing to end-user application:
- Perception Layer — the physical layer of sensors and actuators that interact with the environment. Devices include temperature sensors (DS18B20, ±0.5°C), pressure sensors (BMP388), RFID readers, and actuators such as servo motors and relays. This layer generates raw data in analog or digital form.
- Network Layer — handles transmission of data from devices to processing infrastructure. Protocols include Wi-Fi (IEEE 802.11), Zigbee (250 kbps, 10–100 m), Z-Wave (100 kbps, 30 m), LoRaWAN (0.25–50 kbps, up to 15 km rural), and NB-IoT (200 kbps, LTE cell coverage, deep penetration into buildings).
- Processing Layer — performs computation on raw sensor data. Edge processing occurs on local gateways (e.g., Raspberry Pi 4, NVIDIA Jetson Nano) to reduce latency and bandwidth cost. Cloud processing uses platforms such as AWS IoT Core, Azure IoT Hub, and Google Cloud IoT for storage, analytics, and machine learning pipelines.
- Application Layer — delivers processed data and control interfaces to end users. This includes dashboards (Grafana, AWS IoT SiteWise), mobile apps, and automated control systems. Industrial SCADA systems and consumer smart home apps both operate at this layer.
IoT Communication Protocols
Two protocols dominate IoT messaging at the application layer because of their efficiency on constrained devices:
- MQTT (Message Queuing Telemetry Transport) — a publish-subscribe protocol designed for low-bandwidth, high-latency, or unreliable networks. MQTT v5.0 (OASIS standard, 2019) operates over TCP port 1883 (unencrypted) or 8883 (TLS). Minimum header size is 2 bytes. A device publishes data to a broker (e.g., Mosquitto, AWS IoT Core); subscribers receive messages matching their topic filter. MQTT is used in Amazon Alexa, Facebook Messenger, and industrial sensor networks.
- CoAP (Constrained Application Protocol) — a RESTful protocol for devices too constrained to run HTTP. Operates over UDP port 5683, reducing overhead to a 4-byte fixed header vs. HTTP’s 20+ byte header. CoAP supports GET, POST, PUT, DELETE semantics like HTTP, enabling REST-style IoT APIs on 8-bit microcontrollers with <10 KB RAM. Defined in RFC 7252 (2014).
Consumer IoT Applications
Consumer IoT covers devices in home and personal environments. Three representative categories show the range of sensor, connectivity, and processing requirements:
- Smart thermostat — Ecobee SmartThermostat Premium uses occupancy sensors, Wi-Fi 802.11 b/g/n, and a cloud API to learn schedules and reduce HVAC energy use by up to 26% per Ecobee’s 2022 analysis. Connects to Amazon Alexa, Apple HomeKit, and Google Home.
- Smart security camera — Arlo Ultra 2 streams 4K HDR video over Wi-Fi, performs on-device motion detection using a dedicated ML accelerator, and stores footage to AWS-hosted cloud storage. Active streaming draws approximately 8–10 W.
- Smart speaker — Amazon Echo (5th gen) uses seven microphones with beam-forming to process far-field voice commands. The MediaTek MT8516 quad-core A35 SoC runs Alexa Voice Service. All speech processing occurs server-side over TLS-encrypted WebSocket connections.
Industrial IoT (IIoT) Applications
Industrial IoT applies connected sensors and analytics to manufacturing, logistics, and infrastructure. Two high-value use cases dominate current IIoT deployments:
- Predictive maintenance — vibration sensors (MEMS accelerometers at 10–40 kHz sampling), temperature sensors, and oil quality sensors on rotating equipment (motors, pumps, compressors) stream data to ML models trained to detect bearing degradation signatures. GE’s Predix platform reported a 25% reduction in unplanned downtime in documented deployments. The economic impact of unplanned industrial downtime is estimated at $50 billion per year across Fortune 500 manufacturers (Aberdeen Research, 2019).
- Supply chain tracking — RFID tags (UHF Gen2, ISO 18000-63) and GPS-enabled asset trackers monitor location, temperature (cold chain: 2–8°C for pharmaceuticals), and shock events from manufacturer to end customer. Walmart’s RFID mandate for suppliers in 2022 targets 98%+ inventory accuracy vs. a 63% industry average without RFID.
3 Major IoT Security Risks
IoT security failures are well-documented at scale. Three attack vectors account for the majority of documented IoT compromises:
- Default credentials — devices shipped with factory-set usernames and passwords that users never change. The Mirai botnet (September 2016) exploited this vulnerability across 600,000 devices — routers, IP cameras, DVRs — using a hardcoded list of 62 default credential pairs. The resulting DDoS attack on Dyn DNS on October 21, 2016 took down Twitter, Reddit, GitHub, and Spotify for several hours. Peak traffic reached 1.2 Tbps.
- Unpatched firmware — IoT manufacturers provide infrequent or no OTA (over-the-air) firmware updates. A 2020 analysis by Palo Alto Networks found that 57% of IoT devices were vulnerable to medium- or high-severity attacks and that 98% of IoT device traffic was unencrypted. Devices with no patch mechanism remain exploitable indefinitely after a vulnerability disclosure.
- Lack of encryption — many IoT protocols transmit data in plaintext. MQTT without TLS, Zigbee without AES-128 (which Zigbee optionally supports but not all implementations enable), and older Z-Wave (S0 security with AES-128 CBC but a known IV reuse vulnerability) all permit traffic interception. A 2019 NIST study found only 17% of tested consumer IoT devices enforced TLS for cloud communication.
Consumer IoT vs. Industrial IoT Comparison
Consumer IoT and Industrial IoT share architecture but differ significantly in reliability, security, and latency requirements:
| Dimension | Consumer IoT | Industrial IoT (IIoT) |
|---|---|---|
| Primary protocol | Wi-Fi, Bluetooth, Zigbee | LoRaWAN, NB-IoT, MQTT over LTE, OPC-UA |
| Device cost | $5–$300 | $50–$50,000+ |
| Update frequency | Irregular, user-managed | Scheduled, managed OTA via MDM |
| Uptime requirement | Best-effort, >95% | Mission-critical, >99.9% |
| Latency tolerance | Seconds acceptable | Sub-100 ms for control loops |
| Data volume | KB–MB per day | GB–TB per day per facility |
| Security posture | Frequently default passwords, no TLS | Certificate-based auth, VPN, segmented VLAN |
| Lifespan | 2–5 years | 10–20 years |
Key Takeaways
- There were 15.9 billion IoT-connected devices globally in 2024 per IoT Analytics; 32 billion are projected by 2030.
- IoT architecture has 4 layers: perception (sensors), network (Wi-Fi/Zigbee/LoRaWAN), processing (edge/cloud), and application (dashboards/control).
- MQTT uses a 2-byte minimum header and pub-sub architecture; CoAP uses 4-byte headers over UDP for RESTful constrained device communication.
- The Mirai botnet (2016) compromised 600,000 IoT devices using only 62 default credential pairs and generated a 1.2 Tbps DDoS attack.
- 57% of IoT devices were vulnerable to medium- or high-severity attacks in 2020 per Palo Alto Networks.
- Industrial IoT devices have 10–20 year lifespans vs. 2–5 years for consumer IoT, making firmware update mechanisms critical.
How many IoT devices are there in 2024?
IoT Analytics reported 15.9 billion connected IoT devices globally in 2024. The count excludes smartphones, tablets, and traditional PCs, counting only purpose-built connected devices.
What is MQTT used for in IoT?
MQTT is a lightweight publish-subscribe messaging protocol used in IoT for sending sensor data over low-bandwidth or unreliable networks. Its 2-byte minimum header makes it suitable for microcontrollers.
What caused the Mirai botnet attack?
The Mirai botnet exploited default factory credentials on 600,000 IoT devices in 2016 using a list of 62 username-password pairs, creating a 1.2 Tbps DDoS attack that disrupted major websites.
What are the 4 layers of IoT architecture?
The 4 IoT architecture layers are: perception (sensors/actuators), network (Wi-Fi/Zigbee/LoRaWAN), processing (edge/cloud compute), and application (dashboards/control interfaces).
What is the difference between IoT and IIoT?
IoT refers to consumer-oriented connected devices (smart home, wearables). IIoT is Industrial IoT — connected sensors and systems in manufacturing and logistics requiring >99.9% uptime and 10–20 year device lifespans.
Last Thoughts on the Internet of Things
IoT’s 15.9 billion connected devices in 2024 operate across a 4-layer architecture where the network and processing layers determine latency, cost, and security posture. MQTT and CoAP handle the messaging problem at the application layer for constrained devices. The Mirai botnet demonstrated that deploying devices with default credentials at scale creates attack infrastructure of DDoS magnitude.
With 57% of devices vulnerable and 98% of traffic unencrypted as of 2020, security remains the defining challenge for IoT deployment at enterprise and industrial scale. Industrial deployments address this through certificate-based authentication, VPN tunnels, and VLAN segmentation — controls absent from most consumer IoT devices.
