Smart Places and Infrastructure · Study deck

Smart Home: Design, Energy, and Reliability

A household has chosen devices that can talk to one another, but the first evening scene still fails when the internet drops and wakes people when one sensor is wrong.

Ada is your guide for this deck.

applicationdomainssmart

Ada, the module guide, in a scene from this chapter.

After studying this chapter

Learning objectives

You will be able to:

Explain: A well-designed smart home should have all critical automations (lighting, locks, thermostat schedules) running locally, with cloud used only for remote access and voice assistant NLU processing.
Explain: When setting up a smart home, you must choose between hub-based systems (SmartThings, Hubitat, Home Assistant) and cloud-only systems (Ring, Wyze, TP-Link Kasa).
Explain: Response time: hub-based systems usually respond in 100-500 ms through local processing; cloud-only systems often need 500-2000 ms for cloud round trips.
Explain: Cameras and speakers need Wi-Fi, so they leave the tree at once.

Major section

Incremental Examples

The main review is schedule, manual override, and energy use.

Intermediate Example: A heating scene uses occupancy and temperature.
It must avoid wasting energy when a room is empty and avoid discomfort when a sensor is stale.
It needs false-alarm handling, privacy boundaries, local fallback, and clear user control.

A four-layer false-alarm pipeline addresses pet movement, HVAC currents, weak signal logic, and image verification, reducing weekly false alarms while maintaining true detections.

Major section

Smart Home Architecture

The cloud-assisted route has five stages, from wake-word detection to the bridge and Zigbee bulbs.

The second panel keeps wake-word and AWS processing but replaces two cloud hops, totalling 950 ms, with a 200 ms LAN hop.
That makes the worked total 1,800 ms, a 750 ms saving.

Worked Alexa light command compares a 2,550 ms cloud-assisted route with a 1,800 ms local LAN route, then contrasts Zigbee and Wi-Fi device legs.

Major section

Smart Home Energy Optimization

Security sits above it, where the return is smaller and slower and part of the value is insurance and reassurance rather than a metered saving.

It argues that they are the third purchase, funded by what the base tier already saved.

Smart Home Device ROI Hierarchy: Annual Savings vs. Investment Cost

Major section

Smart Home Energy ROI

The bulbs on the right are the controllable load: they dim, they switch, and they are what a payback sum usually counts.

Key Insight: The smart thermostat alone delivers 54% of total savings, making it the largest-dollar-savings starting point in this household example.

A hub-and-bulb starter set illustrates the investment boundary in the ROI exercise: the controllable load is only part of the cost; the hub that coordinates devices and automation rules also belongs in the system total. Photo: Sho Hashimoto, CC BY 2.0

Major section

Voice Assistant Latency Optimization

Device actuation then crosses the LAN and Zigbee path before audible confirmation arrives at 1000 ms.

The cloud stage consumes 300 ms, and 300 ÷ 1000 × 100 = 30% of the end-to-end delay; it is also the most variable stage.

Voice Command Latency Breakdown: a typical fast round-trip

Major section

Commercial Building Automation

They meet at one controller, whose first job is translation and whose second is scheduling and alarm routing across all four.

Above it sit the outputs a building manager really uses, from an energy dashboard to remote alerts.

Commercial Building Automation System (BAS) Integration

Major section

Smart Home Protocol Comparison

Cameras and speakers need Wi-Fi, so they leave the tree at once.

Everything that remains meets the second question, which asks whether the device must still work when the internet is down.
Critical automation should sit on a branch that survives a cloud outage.

Smart-home protocol comparison by range, response time, and workload

Major section

Smart Homes Can Work Offline

The Belief: If the internet goes down, the entire smart home stops working.

The Reality: It depends entirely on protocol and platform choice.
Thread/Matter devices are designed for local-first operation by default.
A well-designed smart home should have all critical automations (lighting, locks, thermostat schedules) running locally, with cloud used only for remote access and voice assistant NLU processing.

Major section

Hub vs Cloud Smart Homes

When setting up a smart home, you must choose between hub-based systems (SmartThings, Hubitat, Home Assistant) and cloud-only systems (Ring, Wyze, TP-Link Kasa).

This decision affects reliability, privacy, cost, and complexity.
Internet outage: hub-based systems keep local automations working; cloud-only systems may stop working.
Response time: hub-based systems usually respond in 100-500 ms through local processing; cloud-only systems often need 500-2000 ms for cloud round trips.

Major section

Hub vs Cloud Smart Homes (continued)

Privacy: hub-based systems can keep data local unless shared; cloud-only systems send more data to the vendor cloud.

Setup complexity: hub-based systems require moderate hub setup and pairing; cloud-only systems are usually app-first and easier to start.
Device compatibility: hub-based systems can span Zigbee, Z-Wave, and Wi-Fi; cloud-only systems are more limited to a vendor ecosystem.
Monthly cost: hub-based systems are often $0 after hub purchase; cloud-only systems often charge $3-10/month for features.
Technical knowledge: hub-based systems need intermediate routing and IP comfort; cloud-only systems are beginner-friendly.

Deck summary

Key takeaways

The main review is schedule, manual override, and energy use.

The cloud-assisted route has five stages, from wake-word detection to the bridge and Zigbee bulbs.
Security sits above it, where the return is smaller and slower and part of the value is insurance and reassurance rather than a metered saving.
The bulbs on the right are the controllable load: they dim, they switch, and they are what a payback sum usually counts.
Device actuation then crosses the LAN and Zigbee path before audible confirmation arrives at 1000 ms.

Retrieval practice

Recall check 1 of 4

Ada says: answer from memory, then check your reasoning.

Q1A household spends $2,400/year on electricity; HVAC is 45% ($1,080). A $249 thermostat reduces HVAC cost by 18%. In the chapter's gross first-year savings-to-purchase-price comparison, what percentage does the thermostat achieve?

AAbout 78% -- saving $194/year on a $249 investment

BAbout 18% -- matching the energy savings percentage

CAbout 140% -- smart lighting has a higher ROI

DAbout 8% -- barely better than a savings account

Show answer

Answer: A The modeled thermostat saving is 0.18 x $1,080 = $194.40. Gross savings-to-price is $194.40/$249 = 78.1%; net first-year ROI is ($194.40 - $249)/$249 = -21.9%.

Retrieval practice

Recall check 2 of 4

Ada says: answer from memory, then check your reasoning.

Q2A "Good Night" scene controls 16 Zigbee devices and 6 Wi-Fi cloud devices. If each Zigbee device has 99.5% uptime and each Wi-Fi device has 98% uptime, what is the approximate overall scene success rate?

AMultiplying uptime gives about 81.8% scene success.

B99.5% - the average of both protocols

C98.0% - limited by the weakest protocol

D95.0% - a weighted average based on device count

Show answer

Answer: A see answers page

Q3A security system generates 18 false alarms per week. After implementing pet-immune sensors (-45%) and sensor relocation (-20%), how many false alarms remain?

A6.3 per week

B9.9 per week

C11.7 per week

D7.9 per week

Show answer

Answer: D see answers page

Retrieval practice

Recall check 3 of 4

Ada says: answer from memory, then check your reasoning.

Q4For commercial building HVAC optimization, why do buildings with variable occupancy (like universities) see the highest energy savings from occupancy sensors?

AUniversities have older HVAC systems that waste more energy

BStudents generate more body heat than office workers

CPeak-average occupancy gaps create more set-back zones

DUniversity buildings have more windows for natural ventilation

Show answer

Answer: C The gap between peak and average occupancy creates more zones that can be set back Buildings with variable occupancy have the widest gap between "designed-for" capacity and actual usage at any given time.

Retrieval practice

Recall check 4 of 4

Ada says: answer from memory, then check your reasoning.

Q5Complete the smart-home scene reliability calculator:

Asuccess = 1.0

Bsuccess = len(device_reliabilities)

Csuccess = max(device_reliabilities)

Dsuccess = sum(device_reliabilities)

Show answer

Answer: A Scene reliability is the product of each required device's reliability.

Q6A smart home security system generates 15 false motion alerts per week, mostly from pets and sunlight changes. The homeowner starts ignoring all alerts. What is the most effective multi-layer approach to reduce false alarms?

AReplace all motion sensors with cameras and use manual video review for every event

BLayer pet-immune PIR filtering, environmental rules, and AI classification

CIncrease the motion sensor sensitivity threshold to only trigger on large movements

DDisable motion detection during daytime hours when most false alarms occur

Show answer

Answer: B see answers page

Print reference

Answers 1 of 5

Answer key.

A · The modeled thermostat saving is 0.18 x $1,080 = $194.40. Gross savings-to-price is $194.40/$249 = 78.1%; net first-year ROI is ($194.40 - $249)/$249 = -21.9%.

Print reference

Answers 2 of 5

Answer key.

A · 81.8% - the product of all individual reliabilities Scene reliability is calculated as the product of individual device reliabilities. For independent devices: 0.995^16 x 0.98^6 = 0.923 x 0.886 = 0.818 (81.8%). This is why reliability degrades exponentially with device count - each additional device multiplies the failure probability. The key design lesson is to split scenes into "essential" (local-only, high-reliability) and "extended" (cloud-dependent) groups, so critical functions like locking doors always succeed even if non-critical functions like adjusting entertainment systems occasionally fail.

Print reference

Answers 3 of 5

Answer key.

D · 7.9 per week The calculator and pipeline apply each mitigation layer to the alarms that remain after the previous layer. Pet-immune PIR filtering leaves 18 x 0.55 = 9.9 alarms per week. Sensor relocation then removes 20% of that remaining count: 9.9 x 0.80 = 7.92, or about 7.9 false alarms per week. Later layers such as multi-sensor fusion and AI person detection continue from this remaining count.

Print reference

Answers 4 of 5

Answer key.

C · The gap between peak and average occupancy creates more zones that can be set back Buildings with variable occupancy have the widest gap between "designed-for" capacity and actual usage at any given time.
A · Scene reliability is the product of each required device's reliability.

Print reference

Answers 5 of 5

Answer key.

B · Effective false alarm reduction requires layered filtering: hardware-level pet-immune PIR sensors reject small heat signatures, environmental logic rules filter known false-positive patterns (rapid sunlight changes, HVAC airflow), and AI classification can distinguish human movement patterns from pets or environmental noise.