Computer Engineering is coming soon

We're still writing and reviewing the notes for every subject, and we'll open it when every subject is ready. Meanwhile you can practise aptitude, ask Aria about any subject, or see the branches that are ready.

IoT enabling technologies

Concise study notes describing the hardware, communication, power and software technologies that enable IoT systems and the key laws, formulas and example calculations engineers use.

Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

IoT systems are built from many interacting technologies — sensors, radios, processors, power sources, software stacks and cloud services. Engineers use this topic to choose components, design communication links, estimate battery life, and select protocols and platforms (see MQTT, CoAP, Zigbee, LoRaWAN, 6LoWPAN, etc.) when designing working IoT devices and solutions.

Key ideas

  • Sensor: a transducer that converts a physical quantity (temperature, pressure, light, motion, etc.) into an electrical signal; specify range, sensitivity, accuracy, resolution, sampling rate, interface (analog/ADC, I2C, SPI), and power.
  • Microcontroller (MCU): a small computing device with CPU, memory and peripherals designed for embedded control. Assumption: real-time and low-power operation are common priorities in IoT MCUs.
  • System on Chip (SoC): integrates MCU, memory and wireless radios on one chip (reduces BOM and power). Use SoC when space, cost and integrated radio are priorities.
  • Gateway: device that connects local IoT nodes (often using low-power wireless) to the wider network or cloud. Gateways perform protocol translation, buffering and security functions.
  • Firmware: the low-level persistent software running on MCUs/SoCs. It contains device drivers, network stacks and application logic; must be power-aware and robust to resets.
  • PHY (physical layer) and MAC (medium access control): PHY is the hardware+modulation that moves bits on the wire/air; MAC is the protocol that coordinates access (frames, retransmissions). Choose MAC/PHY pair by range, data rate and energy trade-offs.
  • Edge computing: processing data near where it is created (on the device or gateway) to reduce latency, network use and cloud cost.
  • Power management: includes batteries, energy harvesters, PMICs, sleep modes. Key goal: minimize average energy per useful operation (sensing + communication).
  • Security: enabling technologies include secure boot, hardware crypto accelerators, secure elements/TPM for key storage, TLS/DTLS, and OTA update mechanisms for patching; required for device identity, authentication and data confidentiality/integrity.
  • Link budget: accounting of transmit power, antenna gains, path loss and receiver sensitivity to decide whether a communication link will work.
  • Standards and protocols: IoT uses many stacks — constrained devices often use IPv6+6LoWPAN with RPL, CoAP or MQTT/MQTT-SN for messaging, and LoRaWAN/IEEE 802.15.4/Bluetooth Low Energy for radio; cellular IoT options include NB-IoT and LTE-M. (See separate topic notes for each protocol.)

Important formal statements and assumptions:

  • Nyquist Sampling Theorem (band-limited continuous signals): "If a continuous-time signal contains no frequency components above B Hz (it is band-limited to B), then it can be completely recovered from uniform samples taken at a rate fs ≥ 2B." Assumptions: ideal low-pass anti-aliasing filter and infinite-precision sampling. To demonstrate: show that sampling at lower than 2B causes spectral overlap (aliasing) in frequency domain.
  • Shannon–Hartley Capacity Theorem: For an additive white Gaussian noise (AWGN) channel of bandwidth B Hz with signal-to-noise ratio S/N (linear), the maximum achievable error-free information rate C is C = B·log2(1 + S/N) bits/s. Assumptions: AWGN, continuous-time channel, optimal coding over large blocks. Use this to estimate the theoretical upper bound on bit-rate for a given bandwidth and SNR.
  • Friis Transmission Equation (free-space): For line-of-sight free space, received power Pr = Pt·Gt·Gr·(λ/(4πR))^2 where Pt is transmit power, Gt and Gr are linear antenna gains, λ is wavelength and R is distance. Assumptions: far-field (R ≫ λ), no obstacles, matched polarisation.
  • CMOS Dynamic Power: In a switching CMOS digital circuit, dynamic power Pdyn ≈ Ceff·V^2·f where Ceff is the effective switched capacitance, V is supply voltage and f is switching frequency. Assumptions: ignoring leakage and static power; valid when switching dominates.

Drawings to prepare in exam answers (describe exactly):

  • Link-budget diagram: draw transmitter block with Pt (W), antenna gain Gt (dBi), free-space path with distance R labelled, receiver antenna with Gr, and receiver block with Pr and sensitivity. Mark wavelength λ and indicate path-loss term (4πR/λ)^2.
  • Sampling/aliasing diagram: frequency axis showing signal limited to ±B, sampling copies at multiples of fs, illustrate overlap when fs < 2B.
  • Power states diagram: MCU power vs time showing active, sleep, deep-sleep, wake-up; label average current calculation over duty cycle.

Formulas

C = B · log2(1 + S/N) — C: channel capacity in bits per second (b/s); B: bandwidth in Hz; S/N: signal-to-noise ratio (linear). Applies to AWGN channels (Shannon–Hartley). fs ≥ 2 · B — fs: sampling frequency in samples/s; B: highest frequency in the signal in Hz. Applies to ideal band-limited signals (Nyquist). Pr = Pt · Gt · Gr · (λ / (4πR))^2 — Pr: received power in watts (W); Pt: transmit power in W; Gt, Gr: antenna gains (linear, unitless); λ: wavelength in metres (m); R: distance in metres (m). Applies in free-space far-field. Lfs(dB) = 32.44 + 20·log10(f_MHz) + 20·log10(R_km) — free-space path loss with f in MHz and R in km; equivalent to 20·log10(4πR/λ). Pdyn = Ceff · V^2 · f — Pdyn: dynamic power in W; Ceff: switched capacitance in farads (F); V: supply voltage in volts (V); f: switching frequency in Hz. Applies for CMOS switching dominated power. Ebattery (Wh) = Vbattery (V) · Abattery (Ah) — battery energy in watt-hours (Wh); Vbattery in V; Abattery in ampere-hours. Iavg = (Σ Ii · ti) / T — average current over a cycle T with states i of duration ti and currents Ii; use to compute average power Pavg = V · Iavg. Link margin (dB) = Pt(dBm) + Gt(dBi) + Gr(dBi) - Lpath(dB) - Pr_req(dBm) — Lpath: path loss in dB; Pr_req: receiver sensitivity required in dBm. N_floor(dBm) = -174 + 10·log10(B_Hz) + NF(dB) — thermal noise floor at room temperature; NF is receiver noise figure. Sensitivity(dBm) ≈ N_floor(dBm) + SNR_req(dB) — required receiver sensitivity given modulation/coding SNR requirement.

Worked examples

  1. Shannon capacity for a low-power link Given: channel bandwidth B = 200 kHz, SNR = 10 (linear). Find theoretical max capacity. Step 1 (formula): C = B · log2(1 + S/N) Step 2 (substitute): C = 200000 · log2(1 + 10) = 200000 · log2(11) Step 3 (compute): log2(11) ≈ 3.459, so C ≈ 200000 · 3.459 = 691800 b/s Answer: ≈ 692 kb/s (theoretical AWGN upper bound).

  2. Free-space received power (Friis) Given: Pt = 0 dBm (1 mW), Gt = 3 dBi (≈2 linear), Gr = 3 dBi (≈2 linear), frequency f = 915 MHz, distance R = 1000 m. Find Pr in dBm. Step 1: λ = c / f, use c = 3·10^8 m/s → λ ≈ 3·10^8 / 915·10^6 ≈ 0.3279 m. Step 2 (formula linear): Pr = Pt · Gt · Gr · (λ/(4πR))^2. Convert Pt to watts: Pt = 1 mW = 0.001 W. Gt·Gr = 2·2 = 4. Compute (λ/(4πR)) = 0.3279 / (4π·1000) ≈ 2.61·10^-5; square ≈ 6.82·10^-10. Pr = 0.001 · 4 · 6.82·10^-10 = 2.73·10^-12 W. Step 3 (to dBm): Pr(dBm) = 10·log10(Pr in mW) = 10·log10(2.73·10^-9) ≈ -85.6 dBm. Answer: Pr ≈ -85.6 dBm (free-space, ideal polarisation and no obstructions).

  3. Dynamic power and battery life estimate Given: Ceff = 20 pF (20·10^-12 F), V = 3.3 V, f = 1 MHz. Battery: 2000 mAh at 3.7 V. Step 1 (power): Pdyn = Ceff · V^2 · f = 20e-12 · (3.3)^2 · 1e6 = 217.8e-6 W = 217.8 μW. Step 2 (battery energy): Ebatt (Wh) = 3.7 V · 2 Ah = 7.4 Wh = 7.4 W·h. Step 3 (lifetime hours) = Ebatt (Wh) / P (W) = 7.4 / 0.0002178 ≈ 33989 h ≈ 1416 days. Answer: Dynamic switching power ≈ 218 μW; with only this consumption a 2000 mAh battery could theoretically last ≈ 34,000 hours. In real devices radio and other standby currents dominate — always add those when estimating real battery life.

Common mistakes

  • Applying Nyquist fs ≥ 2B to signals that are not band-limited (must apply anti-alias filtering first).
  • Using Friis in obstructed or near-field scenarios. Friis assumes free-space far-field only.
  • Ignoring static/leakage power and radio energy when estimating battery life — counting only MCU dynamic power gives very optimistic results.

For your exam

Types of questions you will get:

  • Definitions and short notes: define sensor, gateway, SoC, edge computing, link budget.
  • Derivations: derive Nyquist sampling requirement and explain aliasing; show Shannon capacity and explain assumptions.
  • Numericals: compute capacity given B and SNR, compute received power using Friis, estimate battery life from power figures.
  • Diagrams: draw and label link-budget, sampling spectrum/aliasing, MCU power-state timeline. What full-marks answers include:
  • Clear definitions with bold term names, units and assumptions stated.
  • Step-by-step derivation or calculation with units converted correctly and final numeric answer shown with correct SI units.
  • Diagrams labelled with all quantities used in calculations and indicating assumptions (e.g., far-field, AWGN, band-limited).

Quick check

  1. State Nyquist sampling theorem in one sentence.
  2. What is the Shannon capacity of a 100 kHz channel with SNR = 100 (linear)?
  3. Write the Friis equation and name two assumptions behind it. Answers: 1) fs ≥ 2B for a band-limited signal — sample at least twice the highest frequency. 2) C = 100000·log2(1+100) = 100000·log2(101) ≈ 100000·6.658 = ~665.8 kb/s. 3) Pr = Pt·Gt·Gr·(λ/(4πR))^2; assumptions: free-space far-field and matched polarisation/no obstacles.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?