By the end of this chapter, you will be able to:
Before diving into this chapter, you should be familiar with:
If you’re deploying hundreds or thousands of IoT sensors across a city, campus, or farm, you face a critical design question: How should your network be structured? LPWAN architectures answer this question differently than traditional networks.
LoRaWAN: Star-of-Stars (DIY Model) Think of LoRaWAN like setting up your own cellular network. You buy gateways (like cell towers) and place them strategically. Sensors send data to any gateway in range—often multiple gateways receive the same transmission (redundancy!). Gateways forward everything to your network server (in the cloud), which deduplicates, decrypts, and routes to your application. You control everything: gateway placement, network configuration, data ownership.
Sigfox: Operator-Managed (Subscription Model) Sigfox is like using Verizon or AT&T for IoT. The Sigfox company operates the base stations—you just buy sensors and subscribe. Your sensors transmit; Sigfox infrastructure receives and forwards data to your server via callbacks (webhooks). Simple, but you’re dependent on Sigfox coverage and subject to their limits (140 messages/day).
NB-IoT/LTE-M: Cellular (Telco Model) Uses existing cell towers from AT&T, Vodafone, T-Mobile, etc. Sensors have SIM cards like phones. Maximum coverage (wherever cell service exists), but higher power consumption and recurring cellular data costs.
| Term | Simple Explanation |
|---|---|
| Star-of-Stars | Devices → Gateways → Network Server (LoRaWAN topology) |
| Gateway | Antenna receiving LoRa transmissions and forwarding to network server |
| Network Server | Central system managing security, routing, deduplication |
| Backhaul | Internet connection from gateway to network server (Ethernet/4G) |
| Operator-Managed | Network run by company (Sigfox)—you just subscribe |
| Private Network | You own and control all infrastructure (LoRaWAN) |
| Callbacks | Webhooks—Sigfox pushes data to your HTTP endpoint |
| Device Class | LoRaWAN A/B/C—determines power use vs downlink capability |
“I need to send data from a farm 10 kilometers away, but there’s no WiFi or cell signal out there!” said Sammy the Sensor.
Max the Microcontroller grinned. “That’s exactly what LPWAN architectures are for! LoRaWAN uses a star-of-stars layout – your sensor talks to a gateway on a hilltop, the gateway forwards to a network server in the cloud. One gateway can cover an entire farm.”
“Sigfox is even simpler,” said Lila the LED. “Your sensor just broadcasts into the air, and any Sigfox base station in range picks it up. No pairing, no handshake. It’s like shouting in a valley – whoever hears you, relays the message.”
Bella the Battery loved the power savings: “NB-IoT piggybacks on existing cell towers, so no new infrastructure needed. And all three architectures are designed for tiny data – a few bytes every hour. That means I can power Sammy for 5 to 10 years on a single battery. The architecture determines the range, the cost, and my battery life!”
Explore Related Learning Resources:
Why These Matter: LPWAN architectures require understanding trade-offs between coverage, cost, and control. The simulations hub helps visualize gateway placement, while the knowledge map shows how architecture choices affect protocol selection and application design.
The Myth: Many assume all LPWAN technologies automatically provide 10+ km range in any environment.
The Reality: Range is highly deployment-specific:
Key Insight: Always conduct a site survey before deploying LPWAN networks. Theoretical range calculations (Friis equation) assume free space—real deployments require accounting for: - Building penetration loss (10-30 dB) - Foliage attenuation (0.5-1 dB per tree) - Interference from other ISM band devices (Wi-Fi, Bluetooth) - Antenna gain and height (every 6 dB doubles range)
Example: A farm deployment claiming “15 km LoRaWAN range” likely uses: - Gateway on 30m tower or silo (height advantage) - High-gain directional antennas (8-12 dBi) - Rural environment with minimal interference - Line-of-sight to most sensors
Replicating this in a city with gateway on a 10m building rooftop? Expect 2-3 km maximum.
Source: NPTEL Internet of Things Course, IIT Kharagpur - This architecture shows how LPWAN technologies fit within the Network Support Technology layer, connecting the Context-Aware Tier (sensors, RFID, data collectors) to the Application Tier (middleware, cloud services, industry applications). LPWAN provides the “low, medium, and high speed communication” bridge in this layered model.
Understanding the differences between LPWAN technologies helps you choose the right solution for your deployment. This comprehensive comparison covers the three major LPWAN options:
| Feature | LoRaWAN | Sigfox | NB-IoT / LTE-M |
|---|---|---|---|
| Deployment Model | Private or operator | Operator-managed only | Telco operator (licensed) |
| Network Ownership | You own (private) or subscribe | Subscription only | Cellular subscription |
| Spectrum | Unlicensed ISM (868/915 MHz) | Unlicensed ISM (868/915 MHz) | Licensed cellular bands |
| Range | 2-15 km (urban/rural) | 10-40 km (excellent) | Cellular coverage (km) |
| Data Rate | 0.3-50 kbps (adaptive) | 100 bps (very limited) | NB-IoT: 250 kbps, LTE-M: 1 Mbps |
| Payload Size | Up to 242 bytes | 12 bytes uplink, 8 bytes downlink | Flexible (1-1000+ bytes) |
| Messages per Day | Unlimited (fair use) | 140 uplink, 4 downlink (hard limit) | Unlimited (data plan limits) |
| Latency | Seconds (Class A), ms (Class C) | Seconds to minutes | <1 second (LTE-M), seconds (NB-IoT) |
| Battery Life | 5-10 years typical | 10-15 years (ultra-low power) | 5-10 years (PSM/eDRX modes) |
| Bi-directional | Yes (Class A/B/C options) | Limited (4 downlinks/day) | Yes (full duplex) |
| Mobility Support | Limited (stationary focus) | Good (handoff between stations) | Excellent (cellular handoff) |
| Infrastructure Cost | $500-1000 per gateway | None (operator-managed) | None (cellular towers exist) |
| Device Cost | $5-20 (modules) | $5-15 (modules) | $10-30 (modules) |
| Setup Complexity | Medium (deploy gateways) | Easy (just subscribe) | Easy (SIM card) |
| Network Scalability | High (add gateways) | Limited (operator capacity) | Very high (cellular infrastructure) |
| Security | AES-128 (end-to-end) | AES-128 (operator-managed) | 3GPP cellular security |
| Geolocation | TDOA (with 3+ gateways) | RSSI-based (built-in) | Cell tower triangulation |
| Standards Body | LoRa Alliance | Sigfox (proprietary) | 3GPP (international standard) |
Choose LoRaWAN when:
Example: Farm monitoring (50-acre property, 100 soil sensors, no cell coverage) → Deploy 2 LoRaWAN gateways
Choose Sigfox when:
Example: Asset tracking (shipping pallets reporting GPS every 30 min = 48 messages/day)
Choose NB-IoT / LTE-M when:
Example: Smart parking meters (city-wide, need reliable uptime, firmware updates, existing AT&T/Verizon coverage)
LoRaWAN (Private Network):
Infrastructure: 10 gateways × $1000 = $10,000
Device modules: 1000 × $10 = $10,000
Network server: $100/month × 60 months = $6,000
Total 5-year cost: $26,000 ($5.20 per device/year)Sigfox (Subscription):
Infrastructure: $0 (operator-managed)
Device modules: 1000 × $10 = $10,000
Subscription: 1000 × $10/year × 5 years = $50,000
Total 5-year cost: $60,000 ($12 per device/year)NB-IoT (Cellular):
Infrastructure: $0 (cellular towers)
Device modules: 1000 × $20 = $20,000
Data plan: 1000 × $5/month × 60 months = $300,000
Total 5-year cost: $320,000 ($64 per device/year)Key Insight: LoRaWAN private networks have higher upfront cost but lowest ongoing cost. Cellular has highest cost but best coverage and reliability. Sigfox is middle ground between DIY and cellular.
A wine cooperative in Napa Valley wants to monitor soil moisture, temperature, and leaf wetness across a 200 km² (20,000-hectare) vineyard region using LoRaWAN. The terrain is gently rolling hills (30 m elevation change), with rows of 2 m tall grapevines. Design the gateway deployment.
Step 1: Establish the link budget.
| Parameter | Value | Source |
|---|---|---|
| Transmit power (ERP) | +14 dBm | US ISM 915 MHz max |
| TX antenna gain | +2 dBi | Sensor whip antenna |
| RX antenna gain | +6 dBi | Gateway fiberglass omni |
| Receiver sensitivity (SF10, 125 kHz) | -134 dBm | SX1276 datasheet |
| Required SNR margin | 10 dB | For 99% reliability |
| Available path loss | 14 + 2 + 6 - (-134) - 10 = 146 dB |
Step 2: Convert path loss to range using the Okumura-Hata rural model.
For 915 MHz at gateway height 15 m (mounted on water tank) and sensor height 1 m:
PL(dB) = 69.55 + 26.16*log(915) - 13.82*log(15) - a(1) + [44.9 - 6.55*log(15)]*log(d)
= 69.55 + 77.5 - 16.3 - 0 + 37.2*log(d)
= 130.75 + 37.2*log(d)
Setting PL = 146 dB:
37.2*log(d) = 15.25
log(d) = 0.41
d = 2.57 kmAdd 6 dB foliage loss (grapevine canopy in summer): effective range drops to ~1.8 km.
Step 3: Calculate coverage area per gateway.
That calculation is for a low-mounted gateway. LoRaWAN gateways on elevated positions achieve significantly more. Re-calculating with the gateway on a 25 m tower (common agricultural silo height):
PL = 69.55 + 77.5 - 13.82*log(25) - 0 + [44.9 - 6.55*log(25)]*log(d)
= 69.55 + 77.5 - 19.3 + 35.7*log(d)
= 127.75 + 35.7*log(d)
Setting PL = 146 - 6 (foliage) = 140 dB:
35.7*log(d) = 12.25
log(d) = 0.343
d = 2.2 kmThe Okumura-Hata model is calibrated for urban/suburban environments. For open agricultural land with line-of-sight from a 25 m tower, empirical LoRaWAN deployments consistently report 3–5 km reliable range at SF10.
Using the empirical value of 3 km (conservative, accounts for foliage and terrain):
Step 4: Factor in gateway redundancy.
LoRaWAN’s strength is multi-gateway reception. For 99.9% uptime, we want every sensor reachable by at least 2 gateways. Overlapping coverage by 30% means:
Step 5: Total deployment cost (5-year TCO).
| Item | Quantity | Unit Cost | Total |
|---|---|---|---|
| Outdoor LoRaWAN gateway (Kerlink/Multitech) | 10 | $1,200 | $12,000 |
| Solar power kit (panel + battery + charge controller) | 10 | $350 | $3,500 |
| Mounting hardware (pole, bracket, lightning arrestor) | 10 | $200 | $2,000 |
| Soil moisture + temperature + leaf wetness sensor nodes | 400 | $45 | $18,000 |
| Network server (TTN/ChirpStack cloud, 5 years) | 1 | $100/mo x 60 | $6,000 |
| Installation labour (2 days, 2 technicians) | 1 | $3,000 | $3,000 |
| Total | $44,500 | ||
| Per sensor per year | $22.25 |
Compare with NB-IoT: 400 sensors × $8/mo cellular plan × 60 months = $192,000 – over 4x more expensive and dependent on cell coverage that may not reach every vineyard row.
The link budget calculation above uses the Okumura-Hata model for path loss estimation. Here’s why those formulas give us the 2.2 km range:
\[\text{PL}(\text{dB}) = 69.55 + 26.16 \log_{10}(f_{\text{MHz}}) - 13.82 \log_{10}(h_b) + \left[44.9 - 6.55 \log_{10}(h_b)\right] \log_{10}(d_{\text{km}})\]
For our 25 m tower at 915 MHz with 140 dB allowable path loss (after foliage margin):
\[140 = 69.55 + 77.5 - 19.3 + 35.7 \log_{10}(d)\] \[\log_{10}(d) = 0.343 \Rightarrow d = 2.2~\text{km}\]
This means each gateway covers \(\pi \times 2.2^2 \approx 15.2~\text{km}^2\) per the model, but empirical agricultural deployments achieve \(28.3~\text{km}^2\) (3 km radius) due to open terrain. Using 30% overlap for redundancy:
\[\text{Gateways} = \frac{200~\text{km}^2}{28.3~\text{km}^2 \times 0.7~\text{per gateway}} \approx 10~\text{gateways}\]
Cost per km² of coverage: \(12,000 / (10 \times 20~\text{km}^2) = \$60/\text{km}^2\) infrastructure investment, amortized over unlimited sensors within that zone.
This gateway placement problem is a specialized version of the sink placement problem in Wireless Sensor Networks. WSN theory says optimal sink (gateway) placement minimizes the maximum hop count and balances traffic load. In LoRaWAN’s single-hop star topology, the “hop count” is always 1, so the problem reduces to geometric coverage: place gateways so every sensor is within radio range of at least k gateways (where k = 2 for redundancy). The vineyard’s rolling terrain makes this harder than flat-earth calculations suggest – always validate with a site survey using a portable LoRa tester before committing to permanent gateway locations.
LoRaWAN uses a star‑of‑stars topology: battery devices send LoRa frames to nearby gateways; gateways forward to a network server, which routes to application servers.
This diagram compares the three LoRaWAN device classes, showing how each trades off power consumption against downlink latency and receive window availability.
Key points: - Uplink frames can be received by multiple gateways; the network server de‑duplicates - Class A/B/C end device classes trade battery vs. latency - Gateways are stateless packet forwarders; backhaul via Ethernet/Wi-Fi/Cellular
Sigfox provides a fully managed network. Devices use ultra‑narrowband to operator base stations; backend routes messages to customer callbacks.
Characteristics: - Very small payloads (uplink‑focused), extremely low power - Coverage and QoS tied to operator footprint - Simple device model (no gateway to manage)
Cellular LPWAN integrates with 4G/5G core. Devices attach to eNodeB/gNodeB; data egresses via EPC/5GC to application servers.
Design notes: - Licensed spectrum, mobility, and SIM/eSIM lifecycle - Power‑saving modes: PSM/eDRX for multi‑year battery life - Best for nationwide coverage and operator‑managed SLAs
Deep Dives:
Comparisons:
Learning:
Explore these AI-generated architectural diagrams that visualize key LPWAN concepts covered in this chapter:
This visualization compares the three major LPWAN technologies, highlighting their positioning in terms of range (2-50 km), data rate (100 bps to 1 Mbps), and battery life (5-20 years).
A structured comparison showing the key trade-offs between unlicensed (LoRaWAN, Sigfox) and licensed (NB-IoT) LPWAN technologies for IoT deployments.
Understanding link budget is essential for LPWAN deployment planning, showing how range is achieved through a combination of transmit power, receiver sensitivity, and spreading factors.
LPWAN network architectures connect to broader networking and IoT deployment concepts:
Architectural Patterns:
Protocol Stack Integration:
Deployment Models:
Technology Trade-offs:
Scalability and Capacity:
LPWAN Technology Deep Dives:
Network Architecture:
Deployment Planning:
Related Technologies:
Learning Resources:
LoRaWAN and Sigfox gateways are radio relay points, not internet gateways. Internet connectivity for devices requires the cloud network server. Local data processing requires separate edge compute infrastructure.
LoRaWAN requires messages to be received by at least one gateway. Coverage holes occur when a gateway fails and there is no overlap. Deploy gateways with 30-40% overlap to ensure coverage redundancy.
LoRaWAN regions have regulatory duty cycle limits (1% in EU868, fair access in US915). Applications that exceed these limits will have messages dropped by the network server without error indication. Calculate and respect air time budgets.
:
This decision flowchart provides an alternative approach to selecting the optimal LPWAN technology based on deployment requirements, ownership model, and technical constraints:
This scatter-plot style diagram visualizes the fundamental trade-offs between power consumption, range, and data rate across LPWAN technologies:
| Chapter | Focus | Why Read It |
|---|---|---|
| LPWAN Comparison and Review | Side-by-side evaluation of LoRaWAN, Sigfox, and NB-IoT across cost, coverage, and capability | Consolidates the architectural differences from this chapter into a practical decision framework |
| LPWAN Assessment: Technology Selection | Structured decision criteria and scoring rubric for choosing between LPWAN options | Apply the trade-off analysis from this chapter to a real deployment scenario |
| LoRaWAN Architecture Deep Dive | LoRaWAN network server functions, ADR, join procedures, and message flows | Go deeper on the star-of-stars internals introduced here |
| NB-IoT Fundamentals | NB-IoT radio access, PSM, eDRX, and 3GPP standards | Understand the cellular core integration described in this chapter |
| LPWAN Fundamentals | Core LPWAN characteristics, link budgets, and spread-spectrum basics | Review the physical-layer foundations that underpin all three architectures |
| WSN Deployment Sizing | Sink placement, coverage planning, and density calculations | The gateway placement maths in this chapter generalises to multi-hop WSN deployments |