Connectivity and Network Design

Introduction

Connectivity is the foundation of every Internet of Things (IoT) system. Sensors, actuators, gateways, edge computers and cloud services can perform useful work only when they exchange information through an appropriate communication network. Network design is therefore not simply a choice between “wired” and “wireless.” It is a structured decision involving distance, data volume, latency, reliability, energy consumption, installation cost, security, topology, spectrum and regulatory constraints.

An industrial temperature sensor, a wearable health monitor, a warehouse asset tag and a connected vehicle have very different communication requirements. Ethernet may be ideal for a fixed camera, Bluetooth Low Energy (BLE) for a battery-powered wearable, and LoRaWAN for widely distributed environmental sensors. Selecting an unsuitable technology can cause short battery life, unreliable control, excessive operating cost or unacceptable security exposure.

This chapter compares major IoT connectivity technologies and develops a practical method for selecting and designing networks.

Learning Objectives

After studying this chapter, the learner should be able to:

1. Explain the roles of wired and wireless connectivity in IoT.
2. Compare Ethernet, Wi-Fi, BLE, Zigbee, Thread, RFID, NFC, LPWAN, LoRaWAN, NB-IoT, LTE-M and 5G.
3. Distinguish bandwidth, range, latency, reliability, topology, energy use and cost.
4. Select an appropriate connectivity technology for a specified IoT application.
5. Identify important security, regulatory, deployment and lifecycle issues.
6. Design a layered connectivity architecture using devices, gateways, networks and cloud services.

1. Connectivity as a Design Decision

A communication technology provides a physical and data-link method for moving information. An IoT solution normally also requires network, transport, application and management functions. For example, a sensor may use BLE to communicate with a smartphone, IPv6 over a local network, and MQTT to publish data to a cloud broker. These are different layers and should not be treated as interchangeable.

Important performance dimensions include:

• Range: maximum practical distance under specified environmental conditions.
• Bandwidth or data rate: amount of data transferred per unit time, normally measured in bit/s.
• Latency: delay between transmission and useful reception.
• Reliability: probability that data arrives correctly and within the required time.
• Energy demand: energy used for transmission, reception, processing and idle operation.
• Topology: physical or logical arrangement, such as star, mesh, point-to-point or cellular.
• Scalability: ability to support more devices, traffic and geographical coverage.
• Cost: device, infrastructure, installation, spectrum, subscription and maintenance cost.
• Security: protection of confidentiality, integrity, identity and availability.

The advertised data rate is not the same as application throughput. Headers, acknowledgements, retransmissions, contention and sleep periods reduce useful capacity. A network should therefore be evaluated using the application requirement rather than a headline specification.

A simple traffic estimate

For a sensor producing a payload of size S bits every T seconds, the average payload rate is:

Here, R_avg is average payload rate in bit/s, S is payload size in bits and T is the reporting interval in seconds. This assumes periodic reporting and excludes protocol overhead. If the device sends 200 bytes every 60 seconds, then S = 1,600 bits and R_avg ≈ 26.7 bit/s. Although this rate is small, the required peak rate, latency and coverage may still rule out some technologies.

2. Wired Connectivity

Ethernet

Ethernet is a mature wired LAN technology used in factories, buildings, gateways, servers, cameras and industrial controllers. It commonly uses twisted-pair copper or fibre-optic cable. Ethernet networks are usually arranged in a switched star: devices connect to switches, while switches may be interconnected hierarchically.

Advantages include high and predictable throughput, low latency, strong physical security, reliable full-duplex operation and the ability to supply power through Power over Ethernet (PoE). PoE can simplify installation by carrying electrical power and data over suitable Ethernet cabling. Fibre provides long distance, high bandwidth and immunity to electromagnetic interference.

Limitations include cabling expense, installation difficulty, physical restrictions and the need for wired access to mobile or difficult-to-reach devices. Ethernet is particularly suitable for fixed equipment, access points, industrial gateways, video systems and building automation controllers.

Other wired considerations

Industrial systems may also use field buses and industrial Ethernet variants. Their value lies in deterministic timing, ruggedness and integration with control equipment. A designer must check cable length, electromagnetic compatibility, grounding, connector protection, redundancy and environmental ratings. In harsh environments, a physically slower but rugged connection may be more suitable than a high-speed office network.

3. Wireless Personal- and Local-Area Technologies

Wi-Fi

Wi-Fi provides wireless LAN connectivity, generally through an access point in a star topology. It offers high data rates and convenient integration with existing IP networks. It is well suited to cameras, displays, appliances, gateways and devices that can tolerate relatively high energy consumption or have mains power.

Wi-Fi performance is affected by distance, walls, interference, channel utilisation, antenna placement and the number of competing clients. Wi-Fi is usually less suitable for tiny battery-powered sensors that transmit only a few bytes because maintaining association and listening for traffic can consume significant energy. Modern power-saving modes improve this situation, but application design remains important.

Bluetooth Low Energy

BLE is designed for short-range, low-power communication. It supports connections between peripherals and central devices, advertising and scanning, and—in appropriate versions and profiles—mesh operation. A wearable may advertise measurements to a smartphone, while a gateway may collect data from many BLE sensors.

BLE is attractive because radio chips are inexpensive, smartphones provide widespread support and devices can sleep between short communication events. Range depends strongly on transmit power, antenna design, data rate and surroundings. BLE is generally used for personal-area sensing, configuration, asset proximity and short-range control rather than wide-area telemetry.

Zigbee

Zigbee is a low-power technology based on IEEE 802.15.4 radios and is widely associated with home and building automation. It supports low-rate devices, coordinators and routers. Mesh routing can extend coverage and improve resilience when powered routers are placed throughout a building.

Zigbee networks require careful channel planning, joining procedures and routing management. Battery devices are normally end devices, while mains-powered devices can act as routers. A mesh may fail or become inefficient if routers are poorly positioned or frequently powered off.

Thread

Thread is a low-power, IPv6-based mesh networking technology built on IEEE 802.15.4. It uses border routers to connect the mesh to other IP networks. Thread is designed for secure, reliable home and building networks and is commonly considered alongside application ecosystems such as Matter.

Compared with a non-IP application network, Thread’s IP foundation can simplify integration with IP-based systems. However, commissioning, border-router availability, radio planning and interoperability still require engineering attention. Thread and Zigbee may use similar radio bands, but they differ in networking architecture and ecosystem assumptions.

4. Identification and Very-Short-Range Technologies

RFID

Radio-frequency identification uses a reader and a tag. Tags may be passive, drawing energy from the reader’s field, or active, using their own battery. RFID is primarily an identification technology rather than a general-purpose sensor network. It is used for inventory, access control, logistics and asset tracking.

Read distance depends on frequency, tag type, reader power, antenna geometry, orientation and nearby materials. Metal and liquid can reduce performance. RFID can identify many objects quickly, but privacy, unauthorised reading and tag cloning must be addressed.

NFC

Near Field Communication is a very short-range technology commonly used for contactless payments, access credentials, device pairing and simple data exchange. Its short range can be a security advantage because deliberate proximity is normally required, but it should not be mistaken for complete security. Relay attacks, malicious tags, insecure applications and stolen credentials remain possible.

NFC is suitable when a user intentionally taps a device. It is not suitable for continuous monitoring across a warehouse or campus.

5. Low-Power Wide-Area Connectivity

LPWAN concept

Low-Power Wide-Area Network (LPWAN) technologies connect small, battery-powered devices over kilometres or large urban areas. They prioritise coverage and energy efficiency over high throughput. Typical applications include metering, agriculture, environmental monitoring, parking and infrastructure condition monitoring.

LPWAN systems commonly send small, infrequent messages. They are a poor choice for high-definition video, continuous audio or highly interactive control. A design must consider message limits, downlink availability, duty-cycle rules, coverage indoors and battery replacement logistics.

LoRaWAN

LoRaWAN is a low-power wide-area networking protocol commonly using LoRa physical-layer modulation in unlicensed spectrum. End devices communicate with gateways, and gateways forward packets to a network server. The architecture is generally star-of-stars rather than a multi-hop mesh: end devices do not normally relay one another’s traffic.

LoRaWAN offers long range, low energy use and flexible private or public deployments. Adaptive Data Rate can optimise data rate and transmit power when radio conditions are known. Limitations include low throughput, variable latency, interference in unlicensed bands, limited downlink capacity and regional regulatory requirements. Device classes provide different receive behaviour: battery-oriented devices listen infrequently, while more responsive classes consume more energy.

NB-IoT

Narrowband IoT is a cellular IoT technology designed for low-rate, wide-area communication, deep coverage and long battery life. It operates within licensed cellular spectrum and benefits from operator-managed infrastructure and authentication. It is suitable for meters and stationary sensors that send modest amounts of data and do not require high mobility or high throughput.

The cellular subscription model and network coverage are advantages, but module and service costs, operator dependence, network availability and potentially higher latency must be considered. Power-saving and extended discontinuous reception modes can substantially reduce energy use, although they also delay downlink communication.

LTE-M

LTE-M, also called eMTC in some contexts, provides more bandwidth and mobility support than NB-IoT. It is better suited to mobile assets, wearables, voice-capable devices and applications requiring more responsive communication. It generally consumes more energy and may cost more than very low-rate alternatives, but it offers a useful balance between cellular coverage and capability.

5G

5G is a family of cellular capabilities rather than a single performance level. It can support enhanced mobile broadband, massive machine-type communication and ultra-reliable low-latency communication, depending on spectrum, deployment and service configuration. Private 5G may provide local control, managed mobility and quality-of-service mechanisms in industrial sites.

5G is not automatically the best choice for every IoT device. Infrastructure, spectrum licensing, subscription, device complexity and energy requirements may be excessive for a soil sensor transmitting once per hour. Its value is strongest where mobility, high capacity, controlled latency, dense device populations or industrial integration justify the investment.

6. Comparative Evaluation

The following qualitative comparison is more useful than treating any single technology as universally superior.

Ranges are indicative, not guarantees. Buildings, antennas, transmit power, regulation and radio interference can change results substantially.

7. Network Topology and Reliability

A star network is simple to manage and allows central policy enforcement, but failure of the central access point can affect many devices. A mesh can provide alternate paths and extend coverage, but routing, commissioning and troubleshooting are more complex. A point-to-point link is simple and predictable but does not scale efficiently. Cellular networks provide broad star-like access through professionally managed infrastructure.

Reliability should be considered at several levels. Radio retransmission may correct occasional errors, but retransmission increases energy use and delay. A gateway may provide local buffering when the cloud is unreachable. Critical control systems may need redundant links, dual power supplies, independent paths and local fail-safe behaviour.

A useful availability model for two independent components in parallel is:

A₁ and A₂ are the individual availabilities expressed as fractions between 0 and 1, and independence is assumed. If two independent links each have availability 0.99, their combined availability is 1 − (0.01 × 0.01) = 0.9999. Real links may share power, towers or backhaul, so independence must not be assumed casually.

8. A Practical Selection Framework

A professional selection process follows these steps:

1. Characterise the application. Record payload size, reporting frequency, peak traffic, downlink needs, latency, mobility and control consequences.
2. Map the physical environment. Identify distance, walls, machinery, electromagnetic interference, hazardous areas, outdoor exposure and available power.
3. Set reliability and availability targets. Decide whether occasional loss is acceptable, whether buffering is possible and whether fail-safe operation is required.
4. Compare candidate technologies. Evaluate coverage, spectrum, topology, installation, device ecosystem, security and lifecycle cost.
5. Plan the complete architecture. Include device firmware, gateways, IP connectivity, application protocols, monitoring and device management.
6. Pilot under realistic conditions. Measure packet loss, latency, battery consumption, coverage and coexistence at representative locations.
7. Design for operation and retirement. Include updates, credential rotation, diagnostics, replacement, data retention and secure decommissioning.

Total cost of ownership includes more than the radio module:

Each term represents a cost over the chosen evaluation period. The expression is a planning framework rather than a universal accounting identity; costs should be discounted or annualised when comparing projects over different time periods.

9. Applied Case: Cold-Chain Monitoring

Background

A food distributor wants to monitor temperature in refrigerated vehicles and storage rooms. Measurements are taken every five minutes. Vehicles move between cities, storage rooms contain metal shelving and refrigeration equipment, and alarms must reach staff promptly.

Problem analysis

A wired Ethernet connection is impractical inside moving vehicles. Wi-Fi may work in depots but would not provide continuous wide-area coverage. BLE sensors are inexpensive and energy efficient, but they require a phone or gateway. LoRaWAN could cover a private depot but would not automatically cover the entire transport route. The system also requires secure device identity, time-stamped records and operation during temporary cloud outages.

Approach and solution

The proposed architecture uses BLE temperature sensors inside each vehicle and storage zone. A vehicle gateway collects BLE data and uses LTE-M for wide-area backhaul. Fixed depot gateways use Ethernet where available and LTE-M as a backup. The gateway stores measurements locally and forwards them when connectivity returns. Alarm logic is duplicated locally so that a dangerous temperature can trigger a local warning even if the cloud is unreachable.

Each sensor is provisioned with a unique identity. Communications are encrypted, gateway firmware is signed, and access credentials are rotated. The pilot measures BLE reliability near metal containers, LTE-M coverage along routes, battery life under the selected reporting interval and alarm delivery time.

Outcome

This hybrid architecture separates short-range sensing from wide-area backhaul. It avoids installing cellular hardware in every low-cost sensor while preserving mobility and operational visibility. The design remains dependent on gateway power, cellular coverage and correct alarm configuration, so maintenance and offline procedures remain essential.

Learning points

• A single technology need not serve every layer of an IoT architecture.
• Gateways can translate protocols, buffer data and enforce local policy.
• Mobility is a major factor in choosing between NB-IoT, LTE-M and local networks.
• Reliability must include offline operation, not only radio link quality.

10. Security, Governance and Regulation

Wireless networks expose signals beyond physical boundaries, while wired networks can still be attacked through ports, switches and compromised devices. Security best practice includes unique device identities, mutual authentication, encryption in transit, secure boot, signed updates, least-privilege access, network segmentation and continuous monitoring.

RFID and NFC systems require particular attention to unauthorised reading, cloning, relay attacks and personal data. Cellular deployments require management of SIM or eSIM credentials and operator relationships. Unlicensed technologies require compliance with regional transmit-power, duty-cycle and spectrum rules. Licensed cellular technologies reduce some interference risks but do not remove cybersecurity responsibilities.

Privacy must be considered when connectivity enables location, behaviour or health monitoring. Organisations should minimise collected data, define retention periods, inform users where required, restrict access and maintain audit records. Safety-critical IoT systems should be assessed using documented risk management, testing and change control.

11. Contemporary Developments and Best Practices

Current IoT design increasingly combines edge computing, IPv6, interoperable application models, device digital identities and automated fleet management. Matter uses IP-based networking across supported local technologies, while private cellular networks are being evaluated for factories, ports and campuses. AI at the edge can reduce communication volume by transmitting events rather than raw data, but it introduces model-management, explainability and update risks.

Best practices are to:

• choose the least complex technology that meets requirements;
• separate sensing, backhaul and application decisions;
• test radio performance at the actual installation site;
• design energy budgets around sleep, transmit, receive and retry states;
• provide local buffering and safe behaviour during outages;
• use standards and documented profiles where interoperability matters;
• monitor battery, signal quality, packet loss and firmware status;
• plan replacement and secure retirement from the beginning.

Chapter Summary

IoT connectivity is a multidimensional engineering decision. Ethernet offers dependable, high-capacity wired communication for fixed equipment, while Wi-Fi provides flexible local networking at higher energy cost. BLE, Zigbee and Thread support low-power local applications, with mesh capabilities useful for building-scale coverage. RFID and NFC are primarily identification and proximity technologies. LPWAN approaches prioritise long range and low energy over throughput; LoRaWAN uses unlicensed spectrum and gateway-based networking, whereas NB-IoT and LTE-M use managed cellular infrastructure. 5G supports demanding industrial, mobile and high-capacity scenarios but may be excessive for simple sensors.

Effective design evaluates range, bandwidth, latency, reliability, topology, energy, security, regulation and total cost together. Hybrid architectures often provide the best result by combining local sensing with gateway processing and wide-area backhaul. Piloting, lifecycle planning, security engineering and offline behaviour are essential professional practices.

Key Terms

Bandwidth: The capacity of a communication channel, usually measured in bit/s.

BLE: Bluetooth Low Energy, a short-range, low-power wireless technology.

Cellular IoT: IoT connectivity provided through mobile network infrastructure.

Ethernet: A widely used wired LAN technology.

Gateway: A device that connects local devices or protocols to another network.

Latency: Time delay between sending and receiving useful information.

LoRaWAN: A low-power wide-area networking protocol commonly using LoRa radio.

LPWAN: Low-Power Wide-Area Network technology for low-rate, long-range devices.

LTE-M: Cellular IoT technology supporting more mobility and throughput than NB-IoT.

Mesh topology: A topology in which nodes may provide multiple communication paths.

NB-IoT: Narrowband cellular technology for low-rate, wide-area IoT communication.

NFC: Near Field Communication for very short-range interaction.

PoE: Power over Ethernet, carrying electrical power and data on Ethernet cabling.

RFID: Radio-frequency identification using readers and tags.

Reliability: The ability to deliver correct service under specified conditions.

Star topology: A network in which devices communicate through a central node.

Thread: An IPv6-based low-power mesh networking technology.

Throughput: Useful application data delivered per unit time.

Topology: The arrangement of nodes and communication links.

5G: A family of cellular technologies supporting high capacity, mobility and advanced IoT services.

Review Questions

Short Answer Questions

1. What factors distinguish bandwidth from throughput?
2. Why is Ethernet often selected for fixed industrial equipment?
3. State two differences between LoRaWAN and cellular IoT.
4. What is the primary purpose of RFID?
5. Why might a gateway be used in an IoT architecture?

Long Answer Questions

1. Compare Wi-Fi, BLE, Zigbee and Thread for a smart-building application.
2. Explain the design trade-offs between NB-IoT, LTE-M and 5G.
3. Describe how topology affects scalability and reliability.
4. Explain the security and privacy risks of RFID, NFC and wireless sensor networks.
5. Develop a systematic connectivity-selection process for a new IoT project.

Analytical / Application Questions

1. A remote water meter sends a small reading every hour and operates on battery power for several years. Compare LoRaWAN, NB-IoT and LTE-M and justify a recommendation.
2. A factory requires mobile robots, video inspection and low-latency control. Propose a connectivity architecture using at least two technologies and explain the role of each.
3. Two independent links each have availability 0.99. Using the parallel-link model, calculate the combined availability and discuss why the result may be optimistic in a real deployment.

Multiple Choice Questions with Answers

Question 1

Which technology is primarily designed for high-capacity wired LAN connectivity?

A. Ethernet

B. NFC

C. RFID

D. LoRaWAN

Correct Answer: A. Ethernet

Question 2

Which topology is most associated with LoRaWAN end devices communicating through gateways?

A. Token ring

B. Star-of-stars

C. Point-to-point mesh only

D. Bus

Correct Answer: B. Star-of-stars

Question 3

Which technology is most suitable for a deliberate tap-to-pay interaction?

A. 5G

B. Ethernet

C. NFC

D. NB-IoT

Correct Answer: C. NFC

Question 4

Compared with NB-IoT, LTE-M generally provides better support for:

A. Passive tags only

B. Mobility and higher throughput

C. Cable-based power delivery

D. Centimetre-range communication only

Correct Answer: B. Mobility and higher throughput

Question 5

A sensor sends 1,600 bits every 60 seconds. What is its average payload rate?

A. 2.67 bit/s

B. 26.7 bit/s

C. 96,000 bit/s

D. 1,660 bit/s

Correct Answer: B. 26.7 bit/s

Question 6

Which technology is based on low-power IPv6 mesh networking?

A. Thread

B. NFC

C. Ethernet

D. RFID

Correct Answer: A. Thread

Question 7

What is a major advantage of a gateway in a hybrid IoT architecture?

A. It eliminates all security requirements

B. It can aggregate local devices and provide protocol or backhaul translation

C. It guarantees unlimited bandwidth

D. It makes sensors independent of power

Correct Answer: B. It can aggregate local devices and provide protocol or backhaul translation

Question 8

Which option is generally most appropriate for a fixed, high-data-rate industrial camera?

A. NFC

B. Passive RFID

C. Ethernet

D. LoRaWAN

Correct Answer: C. Ethernet

Question 9

Under the stated independence assumption, two links each with availability 0.99 have parallel availability of:

A. 0.9801

B. 0.99

C. 0.9999

D. 1.99

Correct Answer: C. 0.9999

Question 10

Which statement best describes LPWAN technologies?

A. They prioritise long range and low energy over high throughput

B. They are designed mainly for high-definition video

C. They always require wired cabling

D. They operate only at centimetre range

Correct Answer: A. They prioritise long range and low energy over high throughput