Skip to main content

Concept

MaaS Blender is designed as an event-driven, discrete-time simulation platform for evaluating Mobility as a Service (MaaS) systems.

Rather than simulating continuous motion or enforcing a global time step, MaaS Blender models system dynamics as a sequence of timestamped events. All changes in system state — such as demand creation, reservation decisions, departures, and arrivals — are expressed explicitly as events occurring at specific simulation times.

Event-Centric Modeling​

In MaaS Blender, events are the primary mechanism for interaction between components.

  • Components typically interact by emitting and receiving events.
  • Each event represents an observable change in the system state.

This approach encourages loose coupling between components.

Separation of Concerns​

The simulation architecture separates responsibilities across independent components, including:

The simulation uses event-centric modeling to separate responsibilities across components. Note that the components shown here are only an example reference implementation. Depending on your use case, you can freely compose them. For example, a mobility service may embed its own routing engine, or a demand generator and user model may be integrated into a single component.

  • The Broker, which orchestrates execution and event propagation
  • Demand Generator, which generate travel demands
  • User models, which model user decision-making
  • Mobility services, which manage vehicles, capacity, and operations
  • Routing and planning components, which propose feasible travel options

Each component is responsible only for its own internal logic and state, and communicates exclusively via events.

Deterministic Execution​

MaaS Blender enforces single-component execution at any point in simulation time. Components are executed sequentially in a deterministic order determined by their next scheduled event time.

As a result, simulation behavior is deterministic given identical inputs and deterministic component implementations. This determinism makes simulation runs reproducible and easier to debug, and allows evaluation results to be consistently interpreted.

Design Goals​

The core design goals of MaaS Blender are:

  • Modularity: enable integration of diverse mobility services and models
  • Determinism: ensure reproducible simulation results
  • Transparency: make system behavior explainable through explicit event traces

These principles guide all architectural decisions described in the following sections.