IoT Value and Product Strategy · Study deck
IoT Evolution: Systems Foundations
Connected systems change when cost, reach, and compute cross a useful threshold.
Ada is your guide for this deck.
After studying this chapter
Learning objectives
You will be able to:
- Distinguish Moore's Law from Dennard Scaling: Explain how physics enabled and then constrained computing, and justify why the distinction matters for IoT device economics
- Apply economic analysis: Evaluate IoT solutions based on computing economics and cost-per-capability trends across technology generations
- Compare centralized vs distributed architectures: Assess the technical and economic trade-offs that favor edge computing in modern IoT deployments
- trace IoT system changes through enabling constraints
Major section
Start With the Story
One more path makes the shift clear.
- A bad remote change could affect many sites.
- Wider reach raised the need for review.
- A service team owned long-term support.
- Online success could not hide these facts.
Major section
Start With the Story (continued)
The story is not just miniaturization; it is the shift from occasional data processing to continuous sensing, local decisions, distributed services, and connected feedback loops.
- Evolution is a set of choices.
- Some jobs need one box.
- Some need a shared service.
- Early control was local.
Major section
In 60 Seconds
An important mid-2000s pivot came as real technologies departed from ideal Dennard scaling and leakage, interconnect, power-density and thermal limits reduced reliance on frequency growth alone.
- Together with continued integration, lower-cost microcontrollers and radios, networking and software-platform advances, this widened the cases in which distributed sensing and edge processing could outperform central-only designs.
Major section
Battery Bruno's Math Bridge: Tiny Antenna, Ten-Year Current Budget
The mathematical gist: At 2.4 GHz, the 12.5 cm wavelength gives a 3.12 cm quarter-wave antenna that fits on a small board.
- A 220 mAh cell, after 1% annual self-discharge for ten years and 20% derating, leaves 159 mAh: only 1.82 microamps average.
- After a 1.0 microamp sleep floor, roughly 0.82 microamps remain for waking and radio work—about 70.6 short reports per day.
Major section
The 10x Technology Cycle
Computing era progression: The following numbers are classroom approximations for comparing orders of magnitude, not market forecasts.
- Economic crossover: A workload that can run across many low-cost edge nodes may beat a central-only architecture on latency, bandwidth, and resilience.
- More transistors therefore do not guarantee a proportionally faster or lower-power IoT device.
- The point is not that every sensor replaces a server; it is that cheap local compute gives architects another placement option.
- Ideal Dennard scaling described how shrinking transistors with lower voltage could keep power density roughly steady; leakage and voltage limits broke that simple relationship.
Major section
Tiny Chips Changed Economics
The important change is not just that devices became smaller.
- Computing also became cheap enough, efficient enough, and connected enough that it started to make economic sense to distribute intelligence across thousands of endpoints instead of concentrating everything in one central server.
Major section
IoT Is a Compute-Placement Shift
Systems evolution is about where useful computation can live.
- In the mainframe era, computation was scarce and centralized.
- For IoT design, the historical pattern becomes a placement checklist.
- A 1960s business could afford only a shared central machine, so applications were organized around scheduled access.
Major section
Choose Placement by Constraint
A practical IoT architecture starts by naming the constraint that decides placement.
- A low-cost MCU is not a finished node.
- The design still needs the sensor, radio, antenna, power regulation, enclosure, mounting, certification, provisioning, and device-management cost.
- The right compute placement is the one that meets the physical constraint with the lowest credible lifecycle risk.
Major section
Distributed Systems Failure Modes
A device can have a local reading, a gateway cache, an MQTT retained message, a cloud device twin, a time-series row, and a dashboard state that do not all update at the same moment.
- The architecture must define which layer is authoritative for commands, telemetry, alarms, and configuration.
- Security state also becomes distributed.
- BLE, Wi-Fi, Thread, LoRaWAN, LTE-M, MQTT, CoAP, OPC UA, and HTTP each carry different assumptions about latency, power, addressing, reliability, and security.
- A sound design includes backpressure, buffering, idempotent commands, timestamp discipline, quality flags, least-privilege credentials, and rollback-capable firmware updates.
Major section
Distributed Systems Failure Modes (continued)
The growing emphasis on performance per watt is visible in the failure model.
- Gateways often translate between constrained field networks and IP services, so they must preserve timestamps, units, quality codes, sequence numbers, and source identity.
- Cloud services receive derived state rather than perfect truth, which means they need late-arrival handling, duplicate suppression, clock-skew tolerance, and explicit command expiry.
- Command path: Separate requested, queued, delivered, applied, rejected, overridden, and expired states.
Major section
Evolution of Internet of Things Systems
The Internet of Things (IoT) has evolved through several distinct phases, reflecting the increasing interconnectedness of devices, people, and systems.
- Each phase represents a significant technological milestone in the journey from simple networks to fully integrated IoT ecosystems.
Deck summary
Key takeaways
One more path makes the shift clear.
- The story is not just miniaturization; it is the shift from occasional data processing to continuous sensing, local decisions, distributed services, and connected feedback loops.
- An important mid-2000s pivot came as real technologies departed from ideal Dennard scaling and leakage, interconnect, power-density and thermal limits reduced reliance on frequency growth alone.
- The mathematical gist: At 2.4 GHz, the 12.5 cm wavelength gives a 3.12 cm quarter-wave antenna that fits on a small board.
- Computing era progression: The following numbers are classroom approximations for comparing orders of magnitude, not market forecasts.
Retrieval practice
Recall check 1 of 2
Ada says: answer from memory, then check your reasoning.
Q1An architect uses the chapter’s era-cost figures to consider edge processing. What conclusion is supported?
Show answer
Answer: A The classroom approximations motivate a design option, not universal replacement of servers.
Q2A designer can now put wireless computation into many ordinary objects. What combination made that deployment plausible?
Show answer
Answer: D The section links distributed endpoints to this combination, not size reduction alone.
Retrieval practice
Recall check 2 of 2
Ada says: answer from memory, then check your reasoning.
Q3Which statement distinguishes Moore's Law from ideal Dennard scaling and explains the IoT power limit?
Show answer
Answer: C Moore’s Law concerns transistor density; Dennard scaling described an ideal power-density relationship that voltage and leakage limits eventually disrupted.
Print reference
Answers
Answer key.
- A · The classroom approximations motivate a design option, not universal replacement of servers.
- D · The section links distributed endpoints to this combination, not size reduction alone.
- C · Moore’s Law concerns transistor density; Dennard scaling described an ideal power-density relationship that voltage and leakage limits eventually disrupted.