Chapters

7 IoT Glossary: G-H

capstone
glossary
g
p

7.1 Overview

This first reference route covers G and H terms and keeps gateway, hardware, web, and hybrid-system definitions together.

This is part 1 of 2. Continue with IoT Glossary: I-P for the second focused route.

7.2 In 60 Seconds

Trace One Message Before Using the Glossary

Picture a greenhouse sensor that reports a dry bed but never starts the pump. The fault may sit in the device, the local link, or the service that reads the message. A glossary helps only when each word points to a real handoff.

Firmware means software stored on a device. A gateway means the bridge between a local device group and a wider network. GPIO means general purpose input and output pins. Inter-integrated circuit (I2C) means a short wired link between chips. An application programming interface means a named way for programs to exchange requests and results; it is called an API. A broker means a service that passes messages between senders and receivers. Telemetry means readings and status sent by a remote device. Message Queuing Telemetry Transport (MQTT) means a lightweight way to send those messages. A payload means the useful data inside one message.

Follow one soil reading through those parts. Unplug the sensor, repeat the message, restart the gateway, and send a value with no time. The pump must stay safe, and the record must show where the handoff failed.

This runway does not replace the definitions below or prove that one design is best. The deeper terms help you name each boundary, compare choices, and record the evidence behind a release decision.

This glossary section covers IoT terminology from G through P, with emphasis on gateway architecture, web and messaging protocols, wireless network choices, hardware interfaces, payload formats, and platform design. The most frequently reused terms in this range include Gateway, MQTT, LPWAN, I2C, OAuth, Platform, Payload, and Publisher.

7.3 Start With The Handoff

Most G-P terms describe a handoff. GPIO and I2C hand a physical signal to firmware. A payload hands program state to a network message. MQTT hands device evidence to subscribers through a broker. OAuth hands a limited permission to an API. A platform hands a prototype to the people who operate, support, and improve it after the demo.

When a term marks a handoff, the useful question is “what could be lost or misunderstood here?” Units can disappear from a payload. A gateway can translate a field incorrectly. A topic can route to the wrong subscriber. A platform can show a polished dashboard while hiding stale data. The term is only trustworthy when the report shows the trace across that handoff.

Read this section with that trace in mind. For each term, look for the artifact on the sending side, the rule at the boundary, and the evidence on the receiving side.

7.4 Learning Objectives

After reviewing this section, you should be able to:

  1. Define key IoT terms from G to P with technical precision.
  2. Differentiate commonly confused concepts such as gateway vs. router, LoRa vs. LoRaWAN, MQTT broker vs. MQTT client, and I2C vs. SPI.
  3. Explain messaging patterns including publishers, subscribers, brokers, topics, and payloads.
  4. Select appropriate protocols, interfaces, and platform patterns for different IoT scenarios.

7.5 G-P Terms Name Handoffs

The G-P range sits in the middle of most IoT designs. Gateway, GPIO, HTTP, I2C, JSON, LoRaWAN, LPWAN, MQTT, OAuth, OPC UA, packet, payload, platform, Protobuf, and publisher are not isolated definitions. They name handoffs between hardware, local networks, application protocols, identity systems, schema owners, cloud services, and people who operate the system after the demo.

Read each term by asking which handoff it describes. GPIO and I2C describe the handoff from a sensor circuit into firmware. Packet, MTU, payload, JSON, and Protobuf describe the handoff from local program state into bytes that can cross a link. Gateway, middleware, microservices, platform, MQTT, HTTP, OPC UA, and OAuth describe the handoff from device or gateway evidence into services with routing, authorization, storage, dashboards, or support ownership.

The main beginner mistake is treating the term as a product label instead of a contract. LoRa is a radio modulation, while LoRaWAN adds join behavior, MAC-layer rules, regional parameters, network-server routing, device classes, and security. MQTT defines publish-subscribe message routing through a broker, but it does not define database retention, dashboard freshness, authorization scope, or command safety. OAuth defines delegated authorization, but it does not replace device identity, certificate handling, or topic ACLs.

HandoffTerms To WatchEvidence To InspectFailure If Misnamed
Sensor to firmwareGPIO, I2C, microcontroller, packetPin mode, pull-up value, bus scan, sample timestampA wiring or bus problem is blamed on the cloud path
Firmware to networkPayload, JSON, Protobuf, MTU, latency, jitterPayload sample, schema version, packet size, retry traceUnits, metadata, or oversized fields disappear in transit
Gateway to serviceGateway, MQTT, HTTP, OAuth, middleware, OPC UATopic, endpoint, scope, translation rule, buffer policyRouting, identity, and offline ownership become unclear
Service to operationPlatform, publisher, subscriber, metadata, microservicesDevice registry, subscriber filter, database row, dashboard ageA polished view hides stale state or missing ownership

Treat each G-P term as a handoff contract: name the boundary, the artifact, and the failure mode before using it in a design claim.

7.6 Prove The Handoff With Traces

In a project review, do not write “the gateway sends data to the platform” without evidence. Split that sentence into inspected artifacts. On the device side, show the sensor value, unit, timestamp, local address, firmware version, and interface evidence. For I2C, that may include the device address, bus speed, pull-up arrangement, and an error or clock-stretching observation. For GPIO, it may include pin mode, voltage level, debounce choice, and board-specific boot pin constraints.

On the message side, show the exact payload and the envelope around it. A payload sample should name fields, units, null behavior, schema version, and what metadata is outside the payload. JSON is readable and easy to debug, but it does not enforce a schema by itself. Protobuf is compact and strongly described by a .proto contract, but it requires schema version discipline and tooling. MTU, packet, latency, and jitter claims should point to a capture, log, or timing sample rather than a single average number.

On the gateway or service side, show routing and identity. For MQTT, record broker address, client id, topic, QoS, retained flag, session behavior, Last Will and Testament if used, subscriber filter, and ACL. For HTTP or HTTPS, record method, path, status code, authentication requirement, retry rule, and error payload. For OAuth, record scopes, token lifetime, refresh flow, and which API enforces the token. For OPC UA or Modbus integration, record the address/register/node mapping and the translation rule that preserves meaning.

On the platform side, name the owner. AWS IoT Core, Azure IoT Hub, ThingsBoard, Eclipse Ditto, InfluxDB, TimescaleDB, Grafana, Node-RED, and custom microservices can all appear in IoT designs, but each owns different states. A platform claim should say whether the service owns device registry, command queue, digital twin or shadow state, storage retention, alert routing, dashboard rendering, analytics jobs, OTA update state, audit logging, or support workflows. If ownership is not named, the glossary term is too broad to support a capstone claim.

A practical review format is four lines: term, handoff, evidence, and consequence. “MQTT” becomes “publisher firmware sends JSON payload to topic site/boiler/temperature with QoS 1; broker log and subscriber insert show duplicate-safe storage; consequence is that subscriber writes must be idempotent.” That sentence is longer than a definition, but it is reviewable.

7.7 Similar Words Carry Different Guarantees

Under the hood, several G-P terms sound interchangeable because they all move data. A packet is a network-layer or link-layer unit with headers and size limits. A payload is the application data carried inside an envelope. Metadata describes context around the payload, such as timestamp, gateway id, quality flag, spreading factor, RSSI, schema version, or tenant. A message may include payload plus headers, topic, properties, and broker state. Mixing those terms makes debugging harder because the missing field may not be in the place the report claims it is.

Gateway, router, middleware, and platform also overlap in everyday speech. A router forwards packets between networks. A gateway may translate protocols, buffer offline telemetry, map local identities to cloud identities, normalize timestamps, run rules, and manage credentials. Middleware may transform messages or coordinate services without owning the whole device lifecycle. A platform may own registry, policy, storage, analytics, command routing, and operational workflows. One physical box or cloud product can contain several roles, so the report should name the role being tested rather than only the product name.

Messaging terms hide guarantee boundaries. An MQTT publisher can send at QoS 0, QoS 1, or QoS 2, but the broker’s acknowledgement is not the same as a database write or operator action. A retained message may be useful for last-known state and dangerous when old state looks current. Wildcard topic filters make subscribers powerful and risky. HTTP status codes can confirm that an API accepted a request, but they do not prove that downstream analytics, alerting, or dashboard refresh completed unless those states are separately inspected.

Radio and timing terms hide physical constraints. LoRaWAN delivery depends on spreading factor, channel plan, regional duty-cycle constraints, gateway placement, join state, receive windows, network-server routing, and backhaul. NB-IoT and LTE-M use licensed cellular networks with different mobility, latency, power, and deployment assumptions from unlicensed LPWANs. Latency averages can hide jitter; jitter can break control loops or ordering assumptions even when average delay looks acceptable. MTU and packet-size limits can turn a neat payload design into fragmentation, truncation, or failed uplinks.

Identity terms hide one final boundary. OAuth authorizes access to APIs, often for user or application delegation. It does not by itself prove that a physical device is genuine, that a gateway certificate is still valid, or that a topic namespace is isolated. HTTPS protects transport, but certificate validation, key storage, refresh behavior, authorization scope, broker ACLs, and audit logs remain separate evidence. The technical term is complete only when the implementation guarantee and its limit are both named.

7.8 Deep Dive: G-P Terms Mark Integration Boundaries

Terms from G through P describe the middle of many IoT systems: the gateway that bridges local devices to wider networks, the hardware interface that collects a reading, the protocol that carries the message, the payload format that represents the value, and the platform that stores or displays the result. Read Figure 7.1 from device stack through gateway translation to the cloud-side contract.

Gateway handoff map showing a device stack, gateway translation and buffering responsibilities, cloud broker or API path, and boundary record fields.
Figure 7.1: G-P terms often describe a handoff where translation, buffering, identity, security, ownership, monitoring, or update responsibility changes.

Follow Figure 7.1 from the device stack through gateway translation and buffering to the broker, API, and cloud evidence path. Each handoff changes identity, protocol, or ownership. A GPIO pin is not an I2C bus; LoRa modulation is not LoRaWAN join and routing; and MQTT publisher, broker, topic, payload, and subscriber remain separate responsibilities even when a demo hides them in one script.

Follow Figure 7.1 from left to right. The device stack owns sensing, the local link, and its initial message evidence. The gateway then translates protocols, buffers outages, maps identity, and creates an upstream contract. The cloud side separates broker or API receipt from registry, subscriber, dashboard, database, and audit outcomes. Each handoff changes what “message,” “payload,” “publisher,” “platform,” and “gateway” mean, so the trace must name the boundary being claimed.

That is why G-P terms should be read as integration terms, not just vocabulary. Gateway is broader than router when it translates, buffers, validates, maps identity, or survives an outage. LoRa is the radio modulation; LoRaWAN adds MAC behavior, join procedure, network server routing, device classes, regional parameters, and security. MQTT is not a database or dashboard; it is a broker-mediated publish-subscribe protocol whose topic and payload choices still need storage, schema, and access-control decisions.

BoundaryTerms That Clarify It
HardwareGPIO, I2C, microcontroller, packet, MTU, and payload terms explain what is sampled, encoded, split, or rejected.
NetworkGateway, LoRa, LoRaWAN, LPWAN, LTE-M, NB-IoT, jitter, and latency terms explain reach, timing, airtime, and backhaul behavior.
ApplicationHTTP, HTTPS, MQTT, OAuth, platform, middleware, microservices, metadata, JSON, Protobuf, and publisher terms explain ownership above the device.

7.8.1 Tie the Term to a Configuration or Trace

Before using a glossary term in a capstone explanation, point to the artifact that proves it. Gateway should connect to a protocol-translation rule, buffering policy, route table, or local queue. MQTT should connect to a broker address, client id, topic name, QoS setting, retained-message policy, Last Will and Testament, or session-expiry setting. OAuth should connect to scopes, token lifetime, refresh behavior, and the API that checks the token.

For hardware and payload terms, check the implementation detail that constrains the claim. I2C depends on bus pull-ups, address conflicts, bus capacitance, and clock speed. MTU depends on the link and headers. JSON, CBOR, and Protobuf change payload size, schema enforcement, debugging, and binary-data handling.

For a gateway claim, collect both sides of the handoff. On the device side, record the source protocol, address, sampling interval, retry counter, local timestamp, and raw or normalized payload. On the gateway side, record translation rules, buffer size, offline retention, credential mapping, topic or endpoint construction, and the rule that decides whether bad data is dropped, quarantined, or forwarded with a quality flag. On the cloud side, record broker ACLs, API route, platform device identity, schema version, database target, subscriber filter, and dashboard freshness rule.

For a messaging claim, write the exact topic, payload sample, QoS level, retained flag, and subscriber behavior. A publisher sends to a topic, but it does not know which subscribers will match the filter. A broker routes messages, but it is not automatically the long-term system of record. A platform may manage device registry, command routing, OTA update state, and analytics, but the glossary term “platform” should be tied to the actual service boundary in the project. For OAuth, record the authorization scope and token lifetime, then separately record how the device or gateway proves its own identity.

  1. Choose the artifact. Example: an ESP32 pinout, I2C bus scan, MQTT packet capture, LoRaWAN uplink metadata row, OAuth consent screen, or platform device registry.
  2. Name the boundary. Example: gateway translation, payload serialization, publisher responsibility, LPWAN airtime, or HTTPS endpoint authentication.
  3. State the consequence. Example: duplicate address, fragmented packet, stale dashboard, rejected token, overloaded broker, missed receive window, or wrong platform owner.

7.8.2 Vocabulary Errors Become Architecture Errors

Mislabeling one G-P term usually moves responsibility to the wrong layer. Calling a gateway a router hides protocol translation, buffering, certificate handling, and offline behavior. Treating a payload as the whole message hides headers, metadata, timestamps, units, quality flags, and schema version. Calling every cloud dashboard a platform hides device provisioning, command routing, identity, storage, analytics, and operations ownership.

Protocol terms also carry hidden contracts. HTTP request-response does not behave like MQTT publish-subscribe. LoRa airtime and LoRaWAN duty-cycle limits do not behave like Wi-Fi throughput. OAuth authorization does not prove device identity by itself. A useful project explanation names the term, the layer, the owner, and the failure mode.

Underneath the wording, each G-P term is a contract with a failure mode. GPIO and I2C failures look like wrong voltage, wrong pin mode, duplicate address, bus capacitance, clock stretching, or missing pull-ups. Packet, MTU, and payload failures look like fragmentation, truncation, oversized JSON, lost metadata, or units that disappear between the firmware and the dashboard. MQTT failures look like duplicate QoS 1 deliveries, retained messages that make old state look current, wildcard subscriptions that receive more than intended, or broker ACLs that let one device publish into another device’s namespace.

Gateway and platform terms are even broader because they combine technical and operational ownership. A gateway may translate, buffer, authenticate, normalize timestamps, run local rules, and upload later. A platform may own provisioning, device twins, shadow state, command queues, storage retention, analytics jobs, alert routing, audit logs, and support workflows. If the glossary term does not name which component owns those states, a capstone design can pass a demo while still failing during a backhaul outage, credential rotation, schema migration, or subscriber-rule change.

For capstone work, use G-P terms to identify who owns the next state transition: the microcontroller firmware, gateway process, broker, network server, middleware service, platform registry, database, dashboard, or support operator.

7.9 How It Works: Turn G-P Terms Into Design Decisions

  1. Start with the physical or protocol artifact: a GPIO pin assignment, I2C address, LoRaWAN join log, MQTT topic, HTTPS endpoint, OAuth scope, payload schema, or platform device registry.
  2. Use the glossary term to name the boundary and its neighboring confusion: gateway vs. router, LoRa vs. LoRaWAN, MQTT broker vs. MQTT client, payload vs. metadata, or platform vs. protocol.
  3. Write the design consequence in one sentence: what fails, retries, queues, fragments, rejects, stores, or alerts when the boundary is wrong?

Beginner Example: An ESP32 room sensor reads an I2C temperature sensor, formats a JSON payload, and publishes it to building/room-12/temperature over MQTT. The key terms are I2C, JSON, payload, publisher, MQTT, and broker.

Intermediate Example: A LoRaWAN soil sensor sends a small uplink through a gateway to a network server. The key terms are LoRa, LoRaWAN, LPWAN, gateway, payload, metadata, latency, and platform; the design consequence is that airtime, spreading factor, gateway placement, and backhaul affect delivery timing.

Advanced Example: An industrial gateway reads Modbus registers, maps them into an OPC-UA information model, serializes selected telemetry with Protobuf, publishes MQTT messages to middleware, and exposes an HTTPS API protected by OAuth scopes. The key question is which component owns translation, authorization, retry, schema evolution, and operator visibility.

7.10 Key Concepts

Gateway: An edge device that aggregates data from local devices and forwards it to cloud or enterprise systems, often handling protocol translation, buffering, and local processing. MQTT (Message Queuing Telemetry Transport): A lightweight publish-subscribe protocol using a broker to route telemetry between devices and applications. LPWAN (Low Power Wide Area Network): A category of wireless networks that trades bandwidth for long range and low power consumption. LoRaWAN: A network protocol built on LoRa modulation, commonly used for long-range sensor deployments. NB-IoT (Narrowband IoT): A licensed cellular LPWAN technology for low-power devices that send small amounts of data. Platform: A complete software environment for device management, data ingestion, storage, analytics, and dashboards. Publisher: A client that sends messages to topics in a publish-subscribe system without knowing which subscribers will receive them.

7.11 For Beginners: IoT Glossary G-P

This section is a quick reference for important IoT words from G to P. Use it when a chapter, project brief, or diagram mentions a term such as “gateway,” “MQTT,” “LPWAN,” or “payload” and you need the definition, plain-language explanation, common mistakes, and related terms in one place.

Navigation

This is part 2 of the IoT Glossary, covering terms G-P. See also:

Glossary A-F - start of the glossary sequence and quick navigation Glossary A-F - ACL through Frame Glossary Q-Z - QoS through 6LoWPAN

Reference Use Metric

Reference usability can be estimated by lookup coverage:

U=RTU = \frac{R}{T}

where RR is resolved lookups and TT is total lookups.

Worked example: If learners perform 220 glossary lookups and 198 are resolved without leaving the section, usability is 198/220=90%198/220=90\%. Improving cross-references and examples should push this ratio higher.


7.12 G

7.12.1 Gateway

Definition: A device that connects networks using different protocols, translating between them and enabling communication between IoT devices and external networks or cloud services.

In simple terms: A translator between different languages - connects your Zigbee sensors to your Wi-Fi network by translating between the two “languages” they speak.

Common confusions:

  • Gateway vs. router - gateways translate protocols, routers only forward packets
  • IoT gateways often combine routing, protocol translation, and edge processing

Related terms: Edge Computing, Broker, Middleware

Synonyms: IoT gateway, protocol converter, edge gateway


7.12.2 GPIO (General Purpose Input/Output)

Definition: Programmable digital pins on a microcontroller that can be configured as either inputs (reading signals) or outputs (sending signals) for interfacing with external components.

In simple terms: The connection points on a microcontroller - like USB ports on a computer, but simpler and more flexible, letting you connect sensors, LEDs, buttons, and other hardware.

Common confusions:

  • GPIO is digital only - use ADC for analog input
  • GPIO current, voltage, pull-up/pull-down mode, and bootstrapping constraints should be checked against the board datasheet to avoid damaged pins or failed boot

Related terms: ADC, DAC, Microcontroller, I2C

Synonyms: digital pins, I/O pins


7.13 H

7.13.1 HTTP (Hypertext Transfer Protocol)

Definition: The foundation protocol of the web, using request-response communication over TCP to transfer hypertext documents, APIs, and other resources between clients and servers.

In simple terms: The language of the web - how your browser asks for web pages and receives them, also used by many IoT devices to send data to cloud services.

Common confusions:

  • HTTP vs. HTTPS - HTTPS adds TLS encryption for security
  • HTTP/1.1 vs. HTTP/2 vs. HTTP/3 have significant performance differences

Related terms: HTTPS, REST, API, CoAP

Synonyms: web protocol


7.13.2 HTTPS (HTTP Secure)

Definition: HTTP protocol secured with TLS encryption, ensuring confidentiality, integrity, and authentication for web communications through cryptographic protection.

In simple terms: Secure HTTP - the same web protocol but wrapped in encryption so nobody can eavesdrop on or tamper with your data in transit.

Common confusions:

  • HTTPS requires certificates and adds computational overhead
  • Not all IoT devices support HTTPS due to resource constraints

Related terms: HTTP, TLS, Certificate, Encryption

Synonyms: HTTP over TLS, secure HTTP


7.14 Continue to Part 2

Continue with IoT Glossary: I-P.