RPL Objective Function Selection Workbench

Compare RPL objective functions, then test why CORPL and acoustic CARP make different route choices

animation

rpl

routing

protocols

6lowpan

objective-function

intermediate

A guided route-choice workbench for RPL OF0, MRHOF, MIN-T, CORPL forwarding sets, and bounded acoustic CARP evidence.

RPL routing Objective functions OF0, MRHOF, and MIN-T Parent choice

Choose the RPL parent by reading the objective-function evidence.

Change the scenario and objective lens to see why the same DIO neighborhood can produce a different preferred parent. The workbench keeps rank, ETX or MIN-T path cost, hysteresis, and loop checks visible together.

MRHOF ETXCurrent selection lens

Parent APreferred parent outcome

ETX 2.5Selected path evidence

Switch allowedHysteresis result

Try

Choose MIN-T, load the parent-flip preset, then change only Path B reverse delivery from 0.800 to 0.300.

Observe

B wins at 1.56 expected transmissions; the reverse-only change raises it to 4.17 and selects A at 2.47.

Explain

Reject “The strongest forward link is the best route”: acknowledgements and control traffic must also survive the reverse direction.

Objective function

The rule set that translates neighbor evidence into rank and preferred-parent decisions inside one RPL Instance.

Rank

A scalar distance from the DODAG root. Rank must progress away from the root so packets can avoid routing loops.

ETX path cost

Expected transmissions can be added along a path. A lower ETX path usually means fewer retransmissions.

Hysteresis

MRHOF can keep the current parent unless the better path improves by enough to avoid constant parent switching.

Candidate parents around node N5

N5 has heard DIO messages from nearby routers. It must choose one preferred parent for upward traffic toward the root.

Root or router Selected parent Rejected candidate Active evidence

Selection rule

Lowest ETX pathMRHOF compares additive path costs and applies hysteresis.

Loop guard

All usable parents progressCandidates must provide forward progress toward the root.

Parent set clue

2 alternatesGood alternates can stay in the parent set even when not preferred.

Candidate-parent table

Compare the evidence each DIO gives N5 before it chooses a preferred parent.

Candidate	Rank evidence	ETX path cost	Guard	Decision

Select a scenario to inspect the parent-choice formula.

Protocol-assumption lens: RPL, CORPL, or acoustic CARP?

Keep one evidence table visible while the protocol assumption changes. RPL chooses a preferred parent; CORPL chooses an ordered forwarding set; CARP chooses an acoustic next hop. They do not share a wire format.

Low-power radio

Try

Load “RPL to CORPL flip,” then change only Protocol lens from RPL MIN-T to CORPL forwarding set.

Observe

The evidence stays fixed: RPL chooses A, while CORPL orders B before A and reports the set benefit separately.

Explain

Reject “The neighbour chosen by RPL is automatically the best CORPL forwarder.” Channel availability changes the explicit objective.

RPL preferred parent AProtocol-specific outcome

Not a forwarding setSet or alternate evidence

Radio parent choiceAvailability or acoustic result

Compare the same entered evidence across the selected bounded score. On narrow screens, scroll sideways for every column.

Candidate	Eligibility	Shared evidence	RPL cost	Decision

C_RPL(j) = advertisedCost_j + 1/(df_j*db_j)

Misconception: “The neighbour chosen by RPL is automatically the best CORPL forwarder.”

RPL chooses A from the fixture evidence. Change only Protocol lens to see whether CORPL keeps that first choice.

Technical boundaries. These are transparent normalized teaching lenses over entered evidence, not normative RPL, CORPL, or CARP implementations. RPL interoperability depends on the deployed objective function, rank rules, metric containers, hysteresis, and loop handling. CORPL also needs forwarding-set signalling, priority, duplicate suppression, channel sensing, and coordination timing. CARP is represented only as rootward progress plus current acoustic delivery and propagation evidence; it does not model sound-speed profiles, Doppler, modem waveforms, mobility, probe aging, collisions, energy, or full protocol messages. Compare protocols only within the displayed medium assumptions and validate with the actual implementation and traces.

OF0

Objective Function Zero seeks a nearby grounded root and uses rank increase from the selected parent. This page uses a fixed rank-stretch teaching model for visibility.

MRHOF

Minimum Rank with Hysteresis Objective Function selects lower additive path cost, commonly ETX, while avoiding excessive churn.

DIO evidence

DIO messages carry DODAG configuration, including the Objective Code Point, rank, and metric-container context used by the instance.

Loop safety

A tempting low-cost neighbor is not usable if it is a descendant, lacks a valid path to the root, or violates implementation constraints.

MIN-T

Each link costs 1 / (d_f × d_b). Add link costs along each usable path; stable path ID breaks an exact tie.

First task

Use the stable scenario with MRHOF.
Switch to OF0 and check whether the parent changes.
Explain why both lenses agree when link quality is similar.

Compare objectives

Open the lossy shortcut scenario.
Predict whether rank or ETX should dominate.
Use the table to defend the selected parent.

Check hysteresis

Open the near-tie scenario.
Move the threshold slider down and up.
Notice when MRHOF holds the current parent.

Technical accuracy notes

Teaching simplification

The OF0 view here uses a fixed `parent rank + 256` stretch so learners can see a hop-like decision. RFC 6552 allows the step of rank to reflect link properties, so OF0 is not always a pure hop-count algorithm.

MRHOF model

The MRHOF view uses additive ETX as the selected metric. The full RFC also defines parent-set behavior, rank conversion, maximum path cost, and several implementation constants.

What rank means

Rank approximates distance from the DODAG root and supports loop avoidance. Exact rank values depend on RPL configuration, objective function, MinHopRankIncrease, and local policy.

MIN-T scope

Technical boundaries. MIN-T assumes stationary independent forward and reverse delivery events and geometric retry-until-success behavior. It does not model estimator confidence, correlated interference, retry caps, queueing, energy, latency, link aging, MAC backoff, multicast, parent churn, or RPL metric-container interoperability. A low cost is routing evidence, not deployment acceptance.

Primary sources and related learning