Sensor Applications · Study deck

Participatory Sensing

Picture residents using their phones to map street noise.

Physics Phoebe is your guide for this deck.

participatory-sensingmobile-sensingcontributor-records

Physics Phoebe, the module guide, in a scene from this chapter.

After studying this chapter

Learning objectives

You will be able to:

Explain: The mathematical gist.: Decibels turn power ratios into additions: $L=10\log_{10}(P/P_{ref})$, so a 4 dB phone-family gap means $10^{4/10}=2.51\times$ acoustic power.
Explain: A repeated +3 dB bias is nearly a doubling, not a rounding error; preserve device provenance and correct against a reference meter rather than blaming ADC bit depth.
Explain: Each contributor brings a phone, a prompt, an access state, a place, a time, and a context note that must be reviewed before records are combined.
Design a contribution prompt that produces reviewable sensor evidence.

Major section

Start With the Crowd Story

Each contributor brings a phone, a prompt, an access state, a place, a time, and a context note that must be reviewed before records are combined.

This check cannot remove every phone difference or participation bias.
People and devices change.
The deeper sections show how contribution records, calibration, privacy, coverage, and feedback support an honest crowd result.

Major section

Phoebe's Field Notes: What a "+3 dB" Phone Bias Really Costs

The mathematical gist.: Decibels turn power ratios into additions: $L=10\log_{10}(P/P_{ref})$, so a 4 dB phone-family gap means $10^{4/10}=2.51\times$ acoustic power.

A repeated +3 dB bias is nearly a doubling, not a rounding error; preserve device provenance and correct against a reference meter rather than blaming ADC bit depth.

Major section

In 60 Seconds

Participatory sensing uses phones carried by contributors to collect observations for a shared sensing task.

The value comes from many scoped contribution records, not from assuming that every phone reading is automatically comparable.
This chapter keeps the workflow bounded.
It teaches how to review contribution evidence before combining phone-based observations.

Major section

Participatory Sensing Workflow

A participatory workflow starts with a shared question and a clear prompt.

The prompt tells contributors what to observe, when to collect, which phone signal or note is needed, and what context must be recorded.
Without that shared structure, the combined data set is difficult to review.

Participatory sensing contribution workflow from shared question through contributor prompt, phone reading, context note, acceptance check, aggregate statement, and retest trigger.

Major section

Contribution Record

A single contribution should be readable without asking the contributor to explain it later.

The timestamp and context note capture when and under what placement, surroundings, or collection posture the record was created.
The same fields should be used for every contributor in the exercise.
Mixed records can still be useful, but the review should identify which fields changed.

Major section

Acceptance Gates

Contributed readings should pass basic gates before they are aggregated.

The goal is not to discard every imperfect record; it is to keep accepted and suspect evidence visibly separate.

Participatory sensing acceptance gates checking prompt match, access state, context note, data quality, comparability, and aggregate-ready status.

Major section

Contribution Modes

The contributor intentionally starts a reading or writes an observation.

It works well for short exercises where each record needs a context note.
The app asks for a contribution when a learning activity reaches a defined point.
It should state the expected phone placement or user action and request only the fields needed for the shared question.

Major section

Contribution Modes (continued)

During review, separate incomplete records from accepted ones rather than filling gaps by assumption.

The app may guide collection using a defined trigger or collection window, but automation still needs a visible review record.
The app must keep stale data separate from current contributions so an old observation cannot satisfy a new prompt.
If the trigger or collection window changes, the review should name that change as a retest condition.

Major section

Aggregation Review

Accepted records supply the aggregate, while suspect or rejected records remain visible beside it rather than disappearing.

The grouping rule—such as a shared prompt, location label, sample window, or context category—states why the selected records are comparable.
That sequence connects the shared result back to the individual contribution evidence.

Participatory sensing aggregation review showing accepted records, suspect records, grouping rule, aggregate statement, scope limit, and retest trigger.

Major section

A Crowd of Phones Is a Crowd of Different Instruments

The application branches explain the attraction; the cross-cutting challenges explain why collection alone is not the engineering result.

This connects the scale promised by crowdsensing to the acceptance and aggregation records developed later in the chapter.

Mobile crowdsensing application map with its cross-cutting review challenges

Major section

Concept Relationships

The shared question defines the contribution prompt, and that prompt selects the phone signal and context fields each record needs.

Access state and data quality then determine whether a contribution is accepted, suspect, or rejected.
Aggregation can combine only comparable accepted records under an explicit rule; it does not repair missing context or incompatible prompts.
Scope limits keep the resulting statement bounded, while retest triggers reopen the workflow when the question, collection window, access path, or acceptance rule changes.

Deck summary

Key takeaways

Each contributor brings a phone, a prompt, an access state, a place, a time, and a context note that must be reviewed before records are combined.

The mathematical gist.: Decibels turn power ratios into additions: $L=10\log_{10}(P/P_{ref})$, so a 4 dB phone-family gap means $10^{4/10}=2.51\times$ acoustic power.
Participatory sensing uses phones carried by contributors to collect observations for a shared sensing task.
A participatory workflow starts with a shared question and a clear prompt.
A single contribution should be readable without asking the contributor to explain it later.

Retrieval practice

Recall check 1 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q1A participatory noise survey receives phone readings with no location context or permission state. How should the dataset treat them?

AMark them suspect until context and access evidence are added.

BAccept them because a larger sample can reduce random variation in the noise estimate

CInclude them in the pooled mean so unusual individual readings have less influence

DReject the whole campaign because one field is missing

Show answer

Answer: A Participatory sensing relies on scoped, comparable contribution records rather than assuming all phone readings are equal.

Retrieval practice

Recall check 2 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q2What is the central data-quality challenge that distinguishes participatory (crowd) sensing from a fixed, professionally calibrated sensor network?

APhones are uncalibrated instruments with different offsets.

BPhones cannot measure anything useful at all.

CThere are always too few contributors to matter.

DFixed networks are actually less accurate than crowds.

Show answer

Answer: A Device heterogeneity and lack of calibration are the defining quality challenge of crowdsensing.

Retrieval practice

Recall check 3 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q3A crowdsensing map averages thousands of contributions but stays biased because a common phone model reads systematically high. Why doesn't adding more contributors fix it?

AThe map needs ten times more contributors and the bias will vanish.

BAveraging increases bias, so they should stop averaging.

CMany contributors share the same device model and thus the same systematic bias.

DThe bias is random noise that just needs more samples.

Show answer

Answer: C Correlated, shared bias does not average away; only per-device calibration or normalization removes it.

Retrieval practice

Recall check 4 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q4A city air-quality map built from volunteer phones shows clean air citywide, but contributions came almost entirely from a few affluent daytime neighbourhoods. What limitation is this, and how should it be handled?

ASensor drift, fixed by recalibrating each phone hourly.

BQuantization error, fixed with a higher-bit ADC.

CNothing is wrong; more contributors always means representative data.

DSpatial-temporal coverage bias from where contributors sampled.

Show answer

Answer: D Uneven contributor coverage biases the aggregate; the fix is to characterize and account for coverage, not to claim citywide validity.

Retrieval practice

Recall check 5 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q5A class collects phone-based sound observations, but some records lack timestamp, access state, and phone placement notes. What is the strongest review feedback before comparing the submissions?

ASeparate incomplete records from accepted records before aggregation

BAverage all records because a larger contribution set cancels out missing evidence

CDiscard the shared prompt because contributor context is enough by itself

DUse the loudest submitted record as the aggregate result

Show answer

Answer: A Participatory sensing records should pass prompt, access, context, and data-quality checks before they are aggregated.

Print reference

Answers 1 of 2

Answer key.

A · Participatory sensing relies on scoped, comparable contribution records rather than assuming all phone readings are equal.
A · Device heterogeneity and lack of calibration are the defining quality challenge of crowdsensing.
C · Correlated, shared bias does not average away; only per-device calibration or normalization removes it.
D · Uneven contributor coverage biases the aggregate; the fix is to characterize and account for coverage, not to claim citywide validity.

Print reference

Answers 2 of 2

Answer key.

A · Participatory sensing records should pass prompt, access, context, and data-quality checks before they are aggregated.