Contents
SpecificationAUTOSAR

AUTOSAR Adaptive Platform

Adaptive Platform execution management, state management, and operating system interface

The Adaptive Platform's Execution Management, State Management, and operating-system interface define how applications on a high-performance ECU are started, stopped, and grouped into states on a POSIX operating system.

What it is

The Adaptive Platform is AUTOSAR’s specification set for software on a high-performance ECU whose application processor has a memory-management unit. Adaptive applications are ordinary operating-system processes. Three specifications define how those processes are created, when they are allowed to run, and what the operating system must provide.

Execution Management is the platform service that starts and stops processes. State Management decides the states of the machine and of its function groups, then asks Execution Management to carry the change out. The operating-system interface records the requirements those services and applications may rely on, including a POSIX PSE51 profile, so one Adaptive stack can run on different operating systems that offer that profile.

These three are the lifecycle backbone of the stack. Communication, logging, diagnostics, and update services all assume that the right processes exist and can be stopped in a defined order.

Who it applies to

Platform and basic-software engineers integrate Execution Management and State Management and show that the chosen operating system meets the interface. Application developers write each process so that it starts, reports that it is running, and shuts down under Execution Management. Architects define machine states and function groups that bundle processes by feature, startup wave, or operating mode.

Safety and cybersecurity work depends on the same model. A safety argument that a function is present in a given mode holds only when the state model starts that function’s processes. A security concept treats Execution Management and State Management as trusted, because they launch processes and accept state-change requests.

What it covers

A machine is one Adaptive Platform instance. A function group is a set of processes started and stopped together. Each function group has named states, and each state lists the processes that should be active. The execution manifest shipped with a process tells Execution Management how to start it, which other processes it depends on, whether to restart it, and which resource group it belongs to. The machine manifest holds machine-level configuration.

Piece Role
State Management Chooses the active machine state and function-group states
Execution Management Creates processes, waits until they report running, and terminates them
Operating-system interface POSIX PSE51 services the platform is allowed to use
Execution manifest Startup options, dependencies, and restart behavior for one process

Execution Management creates each application process, and the process reports its execution state through that interface. Project-specific state-manager logic chooses the target state and requests the transition; Execution Management carries it out. The operating-system document leaves the product choice open and limits the platform and applications to the agreed POSIX subset so behavior stays portable across compliant systems.

How it fits with other documents

Read these specifications with the same Adaptive release: methodology, manifest schemas, and the other foundation clusters. Lifecycle rules from one release and manifests from another are a common source of integration defects.

Communication is a separate cluster. A process uses it after Execution Management has started it and the process has reported that it is running. Update and Configuration Management, Vehicle Update and Configuration Management, Log and Trace, and the Intrusion Detection System Manager are themselves adaptive processes. Execution Management starts them. An update needs State Management to enter a state in which the affected processes can be stopped, replaced, and started again. Where the release provides Platform Health Management, that cluster supervises processes and is the usual partner of Execution Management for restart and recovery.

Classic Platform software schedules runnables inside one ECU image. Adaptive Platform software on this processor schedules POSIX processes under the operating-system interface.

In practice

Group processes that truly start and stop together, such as a driving group, a low-power group, and an update group. Express startup order as manifest dependencies rather than timed delays in application code. Report the running state only when the process can serve its clients, and on shutdown finish persistent writes before exiting.

Keep the state manager as a small project-specific model and test every transition you rely on: cold start, a reduced function group after a fault, entry into an update state, and orderly shutdown. Early on, confirm that the operating system provides the PSE51-based interface required by the chosen release. Re-baseline these specifications and the manifests together whenever the program changes Adaptive release.

What it means for your discipline

  • Base softwareCore

    Platform engineers integrate Execution Management and State Management, map manifests to the selected release, and qualify the operating system against the POSIX PSE51 interface. Process creation, supervision, and recovery sit with this work.

  • HPC & application softwareCore

    Application teams deliver a process plus an execution manifest, report execution state only when the service is ready, and implement the shutdown path Execution Management will run. The process lifecycle stays inside that contract.

  • E/E architectureCore

    Architects own the machine-state and function-group model: which processes exist, which modes start them together, and how driving, degradation, update, and shutdown differ. That model is the integration contract for features on the HPC.

  • Functional safetyPartial

    A safety argument that a function is present in a given mode depends on the state that starts its processes. Startup dependencies, restart after failure, and shutdown order in the manifests belong in that argument.

  • CybersecurityPartial

    Execution Management starts processes with the rights configured for them, and State Management accepts requests that change what is running. Both belong in the trusted computing base and in the security concept.

  • Software update & OTAPartial

    Update and Configuration Management runs as an adaptive process and needs a function-group state in which targets can be stopped and restarted. Update sequences are designed together with State Management.

  • Hardware & siliconPartial

    The operating-system interface assumes an application processor that can host a POSIX PSE51 operating system, with process isolation and enough memory for one process per application. Board and processor choice has to meet that assumption.

  • Diagnostics & lifecyclePartial

    Diagnostic applications and log-and-trace services are processes under Execution Management. A tester only reaches services of processes the active function-group state has started, so diagnostic coverage follows the state model.

No direct stake: Networks & communication