IoT Connectivity and Wireless Networks

Introduction

Connectivity is the enabling layer of the Internet of Things (IoT). Sensors, actuators, embedded controllers, gateways, cloud platforms, and human users can cooperate only when data moves between them with acceptable reliability, latency, energy consumption, cost, and security. IoT connectivity is therefore not synonymous with choosing a radio module. It is a systems-engineering decision involving propagation, spectrum, protocols, topology, power architecture, application requirements, regulation, lifecycle management, and trust.

An IoT network commonly contains several communication domains. A sensor node may use a short-range personal-area network to reach a gateway; the gateway may use Ethernet, cellular, Wi-Fi, or a low-power wide-area network to reach a cloud service; and applications may use IP, MQTT, CoAP, HTTP, or industrial protocols. The best architecture is usually heterogeneous rather than based on one universal technology.

Learning Objectives

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

1. Explain the layered architecture of IoT connectivity and distinguish access, transport, network, and application protocols.
2. Compare major wireless technologies using range, throughput, latency, energy, topology, spectrum, and deployment criteria.
3. Apply link-budget, data-rate, duty-cycle, and energy calculations under stated assumptions.
4. Design a connectivity architecture for a practical IoT deployment.
5. Evaluate interference, scalability, mobility, security, and regulatory constraints.
6. Select appropriate protocols and formulate a validation and lifecycle-management plan.

1. Connectivity as a Multi-Criteria Design Problem

A connectivity choice should begin with requirements, not with a fashionable technology. Important requirements include:

• payload size and message frequency;
• acceptable end-to-end latency and packet-loss probability;
• indoor, outdoor, underground, or industrial propagation conditions;
• node density, mobility, and expected network lifetime;
• battery capacity or energy-harvesting capability;
• ownership of infrastructure and recurring subscription costs;
• security, privacy, availability, and regulatory obligations.

A useful abstraction is the end-to-end path:

Device → local link → gateway or access point → IP transport → IoT service → application

The local link may be Bluetooth Low Energy (BLE), IEEE 802.15.4, Wi-Fi, RFID, or a proprietary sub-GHz radio. The wide-area segment may be Ethernet, fiber, Wi-Fi, cellular, or an LPWAN. Application traffic may use MQTT, CoAP, HTTP, or an industrial protocol. These layers should be evaluated separately because a secure and reliable application can be built over different underlying links.

Figure: A layered IoT connectivity architecture. Gateways may perform protocol translation, buffering, local analytics, and security enforcement.

2. Propagation, Link Budgets, and Capacity

2.1 Radio propagation

A transmitted signal experiences path loss, shadowing, multipath fading, antenna losses, and interference. In free space, the received power is described by the Friis transmission equation:

Here P_t and P_r are transmitted and received powers in watts, G_t and G_r are dimensionless antenna gains, λ is wavelength in metres, and d is separation in metres. The equation assumes unobstructed far-field propagation, aligned polarization, and matched antennas; real deployments often have substantially greater loss.

In decibels, a practical link budget is:

L_p represents propagation loss and L_c represents cable, connector, polarization, and implementation losses. A link is viable only when received power exceeds receiver sensitivity with a margin:

where M is link margin in decibels and S_r is receiver sensitivity in dBm. A positive margin is necessary but not sufficient: interference, fading, regulatory power limits, and antenna detuning may still cause failures.

For free-space propagation, path loss in decibels can be written:

This relation applies under the same free-space assumptions as Friis. Walls, vegetation, vehicles, and human bodies require measurements or an empirically calibrated model.

2.2 Data rate, airtime, and energy

Raw bit rate is not application throughput. Preamble, headers, acknowledgements, inter-frame spacing, retransmissions, coding, contention, and regulatory duty-cycle restrictions reduce useful throughput. If a packet contains L payload bits and has total transmitted length L_f bits at physical rate R_b bits/s, its ideal serialization time is:

The actual airtime also includes protocol overhead and possible waiting time. For a periodically transmitting node, the duty cycle is:

where t_on is time spent transmitting or receiving during period T. This is a design metric and, in some bands and jurisdictions, a regulatory constraint.

Electrical energy is:

where V is voltage in volts, I is current in amperes, and t is time in seconds. For multiple operating states:

This assumes approximately constant voltage and current in each state. Battery life must also account for regulator losses, self-discharge, temperature, pulse capability, ageing, and the difference between nominal and usable capacity.

Worked example. A sensor operates at 3.3 V. During a 40 ms transmission it draws 35 mA, during 10 ms of reception it draws 12 mA, and it sleeps for the remainder of a 60 s cycle at 8 µA. Ignoring regulator losses:

Transmission energy = 3.3 × 0.035 × 0.040 = 0.00462 J.

Reception energy = 3.3 × 0.012 × 0.010 = 0.000396 J.

Sleep energy = 3.3 × 0.000008 × 59.95 = 0.0015827 J.

Total energy per cycle ≈ 0.006599 J. Average power is approximately 0.006599/60 = 0.000110 W, or 110 µW. The result demonstrates why sleep current and wake-up frequency can matter as much as transmit power.

2.3 Shannon capacity and its limits

The theoretical capacity of a bandwidth-limited channel with additive white Gaussian noise is:

C is capacity in bits/s, B is bandwidth in hertz, and S/N is the linear signal-to-noise ratio. The formula assumes the channel model and ideal coding conditions; it is an upper bound, not a guaranteed application rate. Real systems are constrained by modulation, coding, protocol overhead, interference, hardware, and spectrum regulation.

3. Wireless Network Families

3.1 Short-range and personal-area networks

Bluetooth Low Energy is optimized for short bursts, low power, and smartphone or gateway interoperability. It supports advertising and connected operation, device discovery, pairing, characteristic-based data exchange, and, through Bluetooth Mesh, managed many-to-many deployments. It is suitable for wearables, beacons, instruments, and building sensors. Its limitations include shared-spectrum interference, finite range in difficult structures, and dependence on gateways for Internet access.

IEEE 802.15.4-based networks provide low-rate, low-power radio operation and support star or mesh arrangements. Protocol families such as Thread and Zigbee add network and application functions. Mesh routing can extend coverage and provide alternate paths, but routers consume more energy and introduce routing overhead. Thread is IP-based, whereas Zigbee traditionally emphasizes its own application ecosystem, although both can coexist with gateway architectures.

Near-field communication (NFC) works over very short distances and is valuable for provisioning, payment, identification, and secure tap-to-pair interactions. It is not a general substitute for a sensor-area network.

3.2 Wireless local-area networks

Wi-Fi offers high throughput, mature infrastructure, and direct IP connectivity. It is appropriate for cameras, industrial instruments, firmware distribution, and mains-powered devices. Traditional Wi-Fi often has higher energy consumption and contention than low-power mesh technologies, although power-saving modes and newer generations improve efficiency. Capacity planning must consider access-point load, channel reuse, hidden nodes, and uplink congestion.

3.3 Low-power wide-area networks

LPWAN technologies are designed for small payloads, long range, low energy, and low data rates. LoRaWAN uses gateways and a network server with a star-of-stars architecture; end devices communicate over sub-GHz or other regionally permitted bands. Adaptive data rate, spreading factors, gateway diversity, and battery operation are important design concepts. It is unsuitable for continuous high-rate streaming and may be constrained by downlink capacity and regional duty-cycle rules.

NB-IoT and LTE-M use cellular infrastructure. They provide operator-managed coverage, SIM/eSIM identity, mobility-related capabilities, and managed quality-of-service options. NB-IoT targets low data rates and deep coverage, while LTE-M generally offers greater throughput and mobility support. Cellular dependence produces subscription and coverage considerations, and technology sunsets must be included in lifecycle planning.

3.4 Fixed and wired alternatives

Ethernet, industrial Ethernet, fiber, and fieldbus systems remain important. Wired links commonly provide predictable latency, electromagnetic robustness, power delivery, and strong physical security. Wireless should not be selected merely because the project is labelled IoT. A fixed sensor in a factory may be better served by industrial Ethernet or a wired fieldbus.

4. Network Topology and Medium Access

A star topology simplifies management and reduces node forwarding overhead, but the central coordinator is a coverage and availability dependency. A mesh topology provides alternate routes and coverage extension, but requires routing, neighbor maintenance, synchronization, and careful control of latency and energy. A tree topology can simplify addressing but may concentrate failures near higher-level parents. A star-of-stars architecture, common in LPWAN designs, separates end devices from gateway backhaul and network-server functions.

Medium access control determines how nodes share a channel. Carrier-sense multiple access with collision avoidance (CSMA/CA) listens before transmission and uses random backoff; it is effective for moderate, bursty traffic but cannot eliminate collisions or hidden-node problems. Time-division multiple access (TDMA) assigns slots, producing predictable access at the cost of synchronization and scheduling. Frequency-division and code-division methods separate users by frequency or code. Scheduled industrial wireless systems often trade flexibility for deterministic behavior.

Scalability is not simply the number of registered devices. It depends on offered traffic, airtime, retransmission probability, channel occupancy, gateway processing, addressing, routing state, and downlink demand. A network with thousands of sleeping meters may be easier to scale than one with a few hundred cameras.

5. IoT Protocol Stack and Data Semantics

At the link and physical layers, standards define modulation, channels, framing, and access. The network layer commonly uses IPv6, with 6LoWPAN mechanisms enabling compressed IPv6 transport over constrained IEEE 802.15.4 links. Transport may use TCP, which provides reliable ordered streams, or UDP, which provides minimal datagram transport and leaves reliability to higher layers.

At the application layer, MQTT uses a broker and publish/subscribe topics. It decouples publishers from subscribers and supports quality-of-service levels, retained messages, and session-related functions. MQTT is effective for telemetry and event distribution but requires broker availability and careful topic authorization.

CoAP provides a lightweight request/response model over UDP and is designed for constrained devices. It supports methods analogous to web interaction and can be secured using OSCORE or transport protections such as DTLS, depending on architecture. HTTP is widely interoperable but often has greater overhead for highly constrained devices.

Data representation is also a connectivity concern. Compact binary formats reduce airtime, while self-describing formats may simplify integration. Device identity, timestamps, units, calibration status, sequence numbers, and quality indicators should be explicit. Semantic interoperability is not achieved merely by transmitting bytes.

6. Security, Privacy, and Trust

Threats include eavesdropping, traffic analysis, spoofing, replay, rogue gateways, jamming, packet injection, credential extraction, firmware replacement, and denial of service. Security must be layered:

1. Device identity: use unique, provisioned credentials rather than shared default passwords.
2. Link protection: enable authenticated encryption where supported.
3. Network protection: segment devices, restrict routing, and monitor anomalies.
4. Application protection: authenticate clients, authorize topics or resources, validate input, and protect sensitive payloads end to end where necessary.
5. Lifecycle security: support secure boot, signed updates, key rotation, vulnerability response, and secure decommissioning.

Confidentiality, integrity, authenticity, and availability are distinct properties. Encryption alone does not prove that a message came from an authorized device, and authentication alone does not conceal its content. Replay protection generally requires nonces, counters, timestamps, or freshness windows; these mechanisms must handle loss, reset, and clock uncertainty.

Privacy is particularly important for location, health, occupancy, and behavioral data. Data minimization, purpose limitation, retention controls, access logging, and transparent consent should be designed before deployment. Regulatory obligations vary by jurisdiction and sector; engineers must consult applicable radio, cybersecurity, safety, and privacy requirements rather than assuming that a globally available band or library is automatically compliant.

7. Design Methodology

A defensible connectivity design can follow these stages:

Stage 1—Characterize traffic. Specify payload bytes, sampling interval, burst behavior, acknowledgement requirements, and downlink needs.

Stage 2—Characterize the site. Map obstacles, interference, temperature, electromagnetic emissions, antenna locations, and backhaul availability.

Stage 3—Generate alternatives. Compare wired, short-range, mesh, LPWAN, Wi-Fi, and cellular options rather than presuming one technology.

Stage 4—Calculate feasibility. Estimate link margin, airtime, energy, capacity, coverage, and total cost of ownership. Include worst-case rather than nominal conditions.

Stage 5—Prototype and measure. Test packet delivery, latency distributions, current waveforms, roaming, recovery after gateway failure, and behavior under interference.

Stage 6—Secure and operate. Define provisioning, certificates or keys, update mechanisms, monitoring, incident response, and retirement procedures.

Useful performance measures include packet delivery ratio, 95th or 99th percentile latency, energy per successfully delivered payload, availability, mean time to recovery, and cost per connected endpoint. Average latency alone can conceal unacceptable tail behavior.

8. Illustrative Case: Cold-Chain Monitoring

Background. A logistics company must monitor temperature in 1,200 refrigerated containers. Each container records temperature every five minutes, stores readings during coverage loss, and sends an alarm when temperature crosses a threshold. The device should operate for at least two years on a battery and must not rely on continuous high-rate connectivity.

Problem. Containers move between warehouses, ports, and highways. Metal enclosures attenuate radio signals, cellular coverage is variable, and alarms require more urgent delivery than routine telemetry. A solution must also prevent unauthorized modification of temperature records.

Analysis. Wi-Fi would require extensive infrastructure and has unfavorable battery behavior for a mobile asset. A local BLE sensor could reduce sensor energy, but it would require a dependable gateway inside every container. LPWAN could be economical in fixed regions, but roaming and downlink characteristics may be insufficient. Cellular IoT offers wide-area reach, while local storage handles outages.

Approach. An illustrative design uses a low-power temperature sensor, a cellular IoT modem with an eSIM, non-volatile buffering, and a secure element for device credentials. The device sleeps between measurements, transmits periodic summaries, and immediately transmits alarm events with retry and backoff rules. The cloud service authenticates the device, validates sequence numbers, records reception time and sensor time, and alerts operators. Firmware updates are signed and staged; updates are postponed when battery or connectivity conditions are unsafe.

The design team measures antenna performance inside representative containers, evaluates different carrier profiles, tests roaming, and calculates energy using actual modem current waveforms rather than datasheet averages. It also tests a deliberate outage: data should remain locally ordered and be uploaded without duplicate interpretation when coverage returns.

Outcome. The resulting architecture is more resilient than a single-link design because sensing, storage, communication, and application functions are separated. Its costs include cellular subscriptions, certification, secure-element provisioning, and operational monitoring. These are justified only if the value of reduced spoilage and improved compliance exceeds the total lifecycle cost.

Learning points. Mobility changes the connectivity decision; buffering is a reliability mechanism; alarms need different service treatment from periodic telemetry; and security and update capability must be designed into the endpoint.

9. Contemporary Developments and Critical Issues

The boundary between device, gateway, and cloud is becoming more dynamic. Edge computing reduces backhaul and latency by filtering or analyzing data locally. Artificial intelligence can classify vibration, detect anomalies, or predict equipment faults at the edge, but model updates, explainability, energy consumption, and false alarms require governance. AI does not remove the need for reliable sensing and trustworthy time synchronization.

Private cellular networks, software-defined networking, network slicing, Wi-Fi power-saving features, satellite IoT, and non-terrestrial networks expand the design space. Matter seeks greater interoperability in smart-home environments, while industrial deployments increasingly combine time-sensitive networking, deterministic wireless research, and zero-trust security principles. These developments should be assessed against maturity, certification, supply-chain stability, and interoperability—not only advertised performance.

A major research challenge is cross-layer optimization. Increasing spreading factor may improve range but increase airtime and energy. Stronger security may increase computation and packet size but reduce risk. Edge filtering may lower bandwidth while potentially discarding evidence needed for later analysis. Postgraduate engineering requires making these trade-offs explicit, measurable, and reversible where possible.

Best-Practice Checklist

• Begin with quantified application and site requirements.
• Prefer the simplest architecture that meets reliability and security targets.
• Use a link budget, but validate it with field measurements.
• Distinguish nominal throughput from useful application throughput.
• Model peak traffic, retransmissions, downlink demand, and failure recovery.
• Measure energy per successfully delivered message, not only sleep current.
• Design identity, secure updates, logging, and decommissioning from the beginning.
• Provide local buffering and graceful degradation for intermittent connectivity.
• Verify regional spectrum, power, certification, and privacy requirements.
• Plan for technology replacement and operator or vendor discontinuation.

Chapter Summary

IoT connectivity is an end-to-end systems problem involving radio propagation, topology, medium access, protocols, energy, security, regulation, and operations. Link budgets establish physical feasibility, while airtime, duty cycle, capacity, and energy calculations determine whether a design can scale and meet its lifetime target. BLE, IEEE 802.15.4 meshes, Wi-Fi, LPWAN, cellular IoT, Ethernet, and fiber occupy different points in the range–throughput–energy–cost design space. MQTT, CoAP, HTTP, IPv6, and 6LoWPAN provide complementary communication abstractions. Reliable deployments use measurement, buffering, monitoring, authenticated updates, privacy controls, and lifecycle planning. The correct choice is consequently a justified architecture, not merely a selected radio.

Key Terms

IoT connectivity: Exchange of data among IoT devices, networks, platforms, and applications.

Link budget: Accounting of transmitted power, gains, losses, and receiver sensitivity.

Link margin: Difference between received power and receiver sensitivity.

Path loss: Reduction in received signal power caused by propagation.

Airtime: Duration for which a radio channel is occupied by a transmission.

Duty cycle: Fraction of time a transmitter or receiver is active.

LPWAN: Low-power wide-area network designed for small payloads and long range.

Mesh topology: Network in which nodes may forward traffic for other nodes.

Star topology: Network in which endpoints communicate through a central coordinator.

CSMA/CA: Contention-based channel access using carrier sensing and collision avoidance.

TDMA: Channel access method assigning users distinct time slots.

Gateway: Device that connects a local network to another network or protocol domain.

MQTT: Publish/subscribe messaging protocol using a broker.

CoAP: Lightweight web-style protocol for constrained devices.

6LoWPAN: Adaptation mechanisms for IPv6 over low-power wireless personal-area networks.

Receiver sensitivity: Minimum input power needed for specified receiver performance.

Secure boot: Startup process that permits execution only of authenticated software.

Edge computing: Processing performed near data sources rather than exclusively in a remote cloud.

Packet delivery ratio: Successfully received packets divided by transmitted packets.

Review Questions

Short Answer Questions

1. Why is connectivity selection an end-to-end systems-engineering problem?
2. Define link margin and explain why a positive value is necessary.
3. Distinguish star and mesh topologies.
4. What functions does an IoT gateway commonly perform?
5. State two differences between MQTT and CoAP.

Long Answer Questions

1. Compare BLE, IEEE 802.15.4 mesh, Wi-Fi, LPWAN, and cellular IoT for a smart-building deployment.
2. Explain the physical, link, network, transport, and application layers of an IoT architecture.
3. Derive and interpret the practical link-budget equation, including its assumptions and limitations.
4. Discuss IoT security across device, link, network, application, and lifecycle layers.
5. Develop a complete methodology for validating an IoT wireless network before commercial deployment.

Analytical / Application Questions

1. A battery-powered node sends one 80-byte payload every five minutes. Explain which additional parameters are needed to estimate battery life and identify the measurements required in a prototype.
2. A factory has high electromagnetic interference, fixed sensor locations, and strict latency requirements. Evaluate whether wireless is appropriate and recommend an architecture with justification.
3. Design a connectivity architecture for mobile refrigerated assets that must operate during intermittent coverage, distinguish alarm traffic from routine telemetry, and support secure firmware updates.

Multiple Choice Questions with Answers

Question 1

Which expression gives the free-space path loss in decibels?

A. L_FS = 20 log₁₀(4πd/λ)

B. L_FS = B log₂(1 + S/N)

C. L_FS = VIt

D. L_FS = P_t/P_r

Correct Answer: A. L_FS = 20 log₁₀(4πd/λ)

Question 2

A positive link margin means that:

A. The transmitter uses no energy

B. Received power exceeds receiver sensitivity

C. The channel has unlimited capacity

D. Packet collisions are impossible

Correct Answer: B. Received power exceeds receiver sensitivity

Question 3

Which technology is generally most appropriate for high-throughput cameras in a building?

A. NFC

B. LPWAN

C. Wi-Fi

D. Passive RFID

Correct Answer: C. Wi-Fi

Question 4

In MQTT, publishers and subscribers are primarily decoupled by a:

A. Repeater

B. Broker

C. TDMA slot

D. Secure element

Correct Answer: B. Broker

Question 5

If a radio is active for 2 seconds in a 100-second period, its duty cycle is:

A. 0.02%

B. 2%

C. 20%

D. 50%

Correct Answer: B. 2%

Question 6

Which topology allows intermediate nodes to forward packets for other nodes?

A. Mesh

B. Point-to-point only

C. Broadcast-free star

D. Passive bus only

Correct Answer: A. Mesh

Question 7

The Shannon capacity formula is best understood as:

A. A guaranteed application throughput

B. An upper bound under an idealized channel model

C. A battery-life equation

D. A routing algorithm

Correct Answer: B. An upper bound under an idealized channel model

Question 8

Which measure is most directly concerned with energy efficiency of successful communication?

A. Energy per successfully delivered payload

B. Number of device labels

C. Antenna color

D. Topic name length only

Correct Answer: A. Energy per successfully delivered payload

Question 9

Which security mechanism restricts startup to authenticated firmware?

A. Secure boot

B. Carrier sensing

C. Frequency reuse

D. Packet aggregation

Correct Answer: A. Secure boot

Question 10

Why is local buffering valuable in an intermittently connected IoT device?

A. It eliminates all radio interference

B. It preserves measurements during communication outages

C. It guarantees infinite battery life

D. It replaces device authentication

Correct Answer: B. It preserves measurements during communication outages