Energy & Power · Study deck

Energy Cost of Common Operations

Picture a device whose small message triggers a long wake, connection wait, retry, and receive window.

Battery Bruno is your guide for this deck.

awareoperationcosts

Battery Bruno, the module guide, in a scene from this chapter.

After studying this chapter

Learning objectives

You will be able to:

rank CPU, memory, sensing, storage, and radio operations by measured energy cost rather than assumption
build an operation-cost ledger that connects a design change to the dominant energy drain
identify break-even points where local compute stops saving energy against radio transfer
Explain: Counting only the send time hides most of the cost.

Major section

Start With One Expensive Radio Burst

Counting only the send time hides most of the cost.

Tie its version to the exact sleep, wake, sense, compute, store, send, wait, retry, and return sequence.
That one burst can dominate the whole budget.
The goal is to decide when sensing, computing, storing, compressing, or batching saves more energy than it spends.

Operation-cost review starts with scale: small local work can be cheaper than repeating radio sessions, but the product trace still decides the real bottleneck.

Major section

Charging and Communication Energy

The logarithmic ladder then separates arithmetic from data movement: an 8-bit add is inexpensive, while a 32-bit DRAM read can cost orders of magnitude more.

The decision test adds every compute and memory operation, then compares that local total with the measured radio energy avoided on the actual link.

Capacitor voltage rising in four discrete steps with a summed 22.2 nJ loss compared with an 84.5 nJ single jump.

Major section

Bounded Scenario: Remote LoRa Station Cluster

The existing vineyard case already teaches that sensing can create operational value.

Repeating its water-saving story would add little.
The power budget must come from measured normal, retry, join, fault, and cold-temperature traces.
The cost decision is acceptable only if it records who owns each recurring action and what happens when a station is unreachable, a battery misses its service interval, or a gateway credential is suspected compromised.

Major section

Arithmetic, Memory, and Radio Energy Hierarchy

“Compute locally or transmit?” is only answerable after putting arithmetic, data movement, and radio work in the same unit.

Energy per operation is that common unit.
The hierarchy is therefore not simply “integer cheap, floating point expensive.” Once operands travel to distant memory, movement can dominate either arithmetic choice.
A 32-bit SRAM read costs about 50 times the listed 32-bit integer add.

Major section

Arithmetic, Memory, and Radio Energy Hierarchy (continued)

The DRAM read costs about 6,400 adds, or about 173 of the listed 32-bit floating-point multiplies.

The hierarchy is therefore not simply “integer cheap, floating point expensive.” Once operands travel to distant memory, movement can dominate either arithmetic choice.
Local processing is an energy win only when $E_{saved}>0$ and the extra active time, latency, accuracy loss, and memory footprint remain acceptable.
If the same filter spills 2,000 words to DRAM, data movement becomes $2{,}000\times640$ pJ $=1.28\ \mu$ J and reverses the result.

Sensor, ADC, optional compute, and radio path above a logarithmic operation-energy ladder.

Major section

Summary

Simple local work is often worthwhile when it removes radio sessions, payload, storage, or sensor activity that the service does not need.

Radio cost includes startup, listening, joins, acknowledgments, retries, payload, and shutdown, not only the bytes transmitted.
Batching, filtering, aggregation, compression, and offloading must preserve latency, freshness, loss, alarm, and evidence requirements.
The final proof is measured charge per useful report before and after the change.

Deck summary

Key takeaways

Counting only the send time hides most of the cost.

The logarithmic ladder then separates arithmetic from data movement: an 8-bit add is inexpensive, while a 32-bit DRAM read can cost orders of magnitude more.
The existing vineyard case already teaches that sensing can create operational value.
“Compute locally or transmit?” is only answerable after putting arithmetic, data movement, and radio work in the same unit.
The DRAM read costs about 6,400 adds, or about 173 of the listed 32-bit floating-point multiplies.

Retrieval practice

Recall check 1 of 3

Battery Bruno says: answer from memory, then check your reasoning.

Q1A developer wants to save energy by rewriting a small temperature filter, but the measured trace shows most charge is spent on radio startup, receive windows, and retries. What is the best review finding?

ARewrite the small filter first because CPU work dominates the trace.

BTransmit more often to shrink each report because shorter reports offset the additional radio setups.

CTarget radio startup, windows, batching, and retry behavior first.

DIgnore retries because only payload bytes cost radio energy.

Show answer

Answer: C Energy optimization should start with the largest measured contributor in the whole duty cycle.

Retrieval practice

Recall check 2 of 3

Battery Bruno says: answer from memory, then check your reasoning.

Q2A team batches sensor readings so the radio wakes less often, but the batch can now hold an alarm for several minutes. What should the practitioner review require?

AA service-constraint check that separates delay-tolerant data from alarms.

BApprove the change because fewer radio starts reduce session overhead.

CAdd more flash logging by default because writes and resets have negligible energy cost.

DRemove local filtering so the batch retains the original sensor detail.

Show answer

Answer: A The ledger has to protect the service contract while proving the energy effect with before and after traces.

Retrieval practice

Recall check 3 of 3

Battery Bruno says: answer from memory, then check your reasoning.

Q3A compression routine makes payloads smaller but keeps the device awake longer and does not reduce radio retries in weak-signal tests. What is the strongest conclusion?

ACompression saves energy whenever payload byte count falls.

BCompression needs proof that charge per useful report declines.

CWeak-link retries can be ignored as inconvenient test noise.

DRaw samples should be stored in flash to improve compression.

Show answer

Answer: B A local optimization must reduce measured whole-cycle charge while preserving the service contract.

Print reference

Answers

Answer key.

C · Energy optimization should start with the largest measured contributor in the whole duty cycle.
A · The ledger has to protect the service contract while proving the energy effect with before and after traces.
B · A local optimization must reduce measured whole-cycle charge while preserving the service contract.