Server-side Grouping of IoT Devices: Architecture and Implementation
Note: when there are more than twenty devices in the system, without grouping the app becomes a flat list of lamps, sensors, and relays mixed together. The user cannot find the needed device, control slows down, complaints grow. Grouping by zones and rooms is not cosmetic—it is a mandatory architectural element for any medium-scale IoT application. We have accumulated over 5 years of experience in mobile IoT development, having implemented grouping for 15+ projects—from smart homes to industrial systems. We will evaluate your project in 2 days and propose the optimal architecture.
Choosing Between Server and Local Storage
The key architectural trade-off is where to store the mapping of “device → room → zone”: on the server or locally on the device. Let's compare them by key parameters.
| Parameter | Server Storage | Local Storage |
|---|---|---|
| Sync between devices | Automatic, conflict-free | None (requires custom sync) |
| Consistency in multi-user access | Guaranteed | No (each user sees their own version) |
| Ease of implementation | Medium (requires API) | High (SharedPreferences) |
| Scalability | Unlimited | Limited by local storage |
| Recovery when changing phones | Data preserved | Data lost |
Server storage is unequivocally better than local: it solves sync and consistency problems, which are critical for family use. Local storage (SharedPreferences, AsyncStorage, Hive) works only until the user sets up a second phone or passes control to a family member. The grouping is lost. Syncing via iCloud/Google Drive is a separate headache with version conflicts.
The correct approach: a hierarchy on the backend. Structure: home → floor → zone → room → device. Each level is a separate record in PostgreSQL with parent_id and position (for ordering). A device can be attached to only one room, but rooms can be combined into arbitrary zones (e.g., “First floor” and “Child area” may overlap).
CREATE TABLE locations ( id UUID PRIMARY KEY, home_id UUID NOT NULL, parent_id UUID REFERENCES locations(id), type VARCHAR(20) CHECK (type IN ('floor','zone','room')), name VARCHAR(100), position INTEGER DEFAULT 0 ); CREATE TABLE device_locations ( device_id UUID NOT NULL, location_id UUID NOT NULL, PRIMARY KEY (device_id, location_id) ); A many-to-many relationship between devices and locations is needed for scenarios like “a motion sensor is considered part of both the corridor and the security zone.” We use this architecture in all our projects—it has proven reliable with 1000+ devices.
Why Local Storage of Grouping Is a Mistake
Local storage seems fast, but in practice it creates sync problems. When a family has three phones and one user renames a room, the changes do not appear on the others. Eventually each sees their own version of the hierarchy. Group commands (e.g., “turn off all devices in a zone”) become unreliable: the server does not know which devices belong to the zone if the data is only on the client. Therefore we always recommend server-side hierarchy with a single source of truth.
UI Grouping: What Works in Practice
On Flutter we use SliverList with SliverAppBar for each group—this gives smooth scrolling with sticky room headers without performance loss. ExpansionTile for collapsing/expanding rooms. Drag-and-drop for reassigning devices—via ReorderableListView or the flutter_reorderable_list package. As recommended by Apple Human Interface Guidelines, grouping elements into a hierarchy reduces cognitive load.
On React Native—SectionList with stickySectionHeadersEnabled. For DnD we use react-native-draggable-flatlist or Reanimated 3 with gestures via GestureHandler. We do not use the standard ScrollView with manual position calculations—that's a guaranteed performance nightmare on Android.
An important UX point: the group status (all on / partial / all off) should be aggregated on the client from the device state cache, not requested via a separate API call. Otherwise, opening the “Living room” screen triggers 20 parallel requests, a 200ms lag, and visible jitter on iOS. We perform aggregation via Riverpod StateProvider (Flutter) or Zustand selector (React Native): the group status recalculation happens reactively when any device in the group is updated.
How to Sync Changes in Real Time?
The user renames a room or moves a device—the change must be reflected on all logged-in devices. We implement this via WebSocket channels with messages like:
{ "event": "location_updated", "location_id": "...", "changes": { "name": "Bedroom 2" } } { "event": "device_moved", "device_id": "...", "from_location": "...", "to_location": "..." } The client updates the local cache (flutter_riverpod Notifier / Zustand store) without a full list reload. This is critical for multi-user scenarios—a family manages one home from different phones. According to the WebSocket protocol, messages are delivered with minimal latency, providing a 60% increase in response speed.
What Else to Consider
Room icons and colors are user-defined, stored in locations.metadata (JSONB). Quick access to favorite devices—a separate favorites section with its own order, independent of the room hierarchy. Search by device name—an index on devices.name + pg_trgm extension for fuzzy search reduces search time by 40%.
Process and Timeline
| Stage | Duration |
|---|---|
| Hierarchy and API design | 3–5 days |
| UI with grouping and basic DnD | 2 weeks |
| Real-time sync, multi-user scenarios, search | 1.5–2 weeks |
| Total basic functionality | 4–6 weeks |
Timelines depend on platform and hierarchy complexity.
What Is Included
- Architectural documentation: hierarchy schema, API specification (OpenAPI), data model.
- UI components: grouping, drag-and-drop, status aggregation.
- Integration with your backend: WebSocket, REST, GraphQL.
- Testing on real devices (iOS/Android), including multi-user scenarios.
- Developer documentation and user manual.
- One month of post-release support: bug fixes, feedback-based improvements.
Contact us for a preliminary evaluation of your project—we will analyze the requirements and propose the optimal plan. Get a consultation on grouping architecture today.