Secure your vehicle's "black box" data during sudden power loss with Emergency Storage Protocol

In the real world, an electric vehicle can lose all power in an instant. A traffic accident, a severe water incident, a catastrophic hardware failure—any of these can trigger an emergency shutdown of the vehicle's domain controllers. In that moment, there is a time window of just one to two seconds, often even less, to perform the most critical data preservation tasks.

Whether the vehicle’s complete operational data and software logs can be saved to disk in this race against time is of paramount importance. The ability to secure this high-fidelity data and support its emergency upload directly impacts accident reconstruction, liability assessment, and even future safety optimizations and system upgrades.

The Demand and the Dilemma

It's well known that a modern electric vehicle is an intelligent electronic system defined by its software. During normal operation, it generates a massive amount of critical data: bus signals, DTC fault codes, software logs, Ethernet packets, and environmental and autonomous driving perception data. This raw data is typically gathered and stored by an on-vehicle collection module, like the EXD On-Vehicle Data Foundation, before being uploaded to the cloud.

Because this data is generated at a millisecond level, a standard storage strategy is used to balance performance with the lifespan of the storage hardware: data is cached in high-speed memory and written to the persistent disk periodically in batches.

This standard procedure, however, has an Achilles' heel: a sudden, total loss of power. The periodic write-to-disk strategy is no match for an emergency shutdown. This demands a specialized storage mode that can:

1. Instantly recognize that a critical event is occurring.

2. Trigger a "race against the clock" emergency storage and transmission mechanism.

3. Guarantee that every last second of incident data is saved, complete, unalterable, and transmittable.

Crucially, if the vehicle stabilizes and avoids a complete shutdown, this emergency mode must be smart enough to automatically revert to its normal, periodic storage routine to prevent excessive wear and tear on the storage hardware.

The Solution Starts with Instantaneous Detection

Identifying the onset of a critical event is the first and most vital link in the data-preservation chain. With a 1-2 second window, relying on a round trip to the cloud for analysis is simply not an option.

For this reason, the EXD Data Foundation employs an intelligent edge-first approach: Real-Time Edge Computing + a Triggered Emergency Storage Mode. This means that when an algorithm running directly on the vehicle's hardware detects a critical event—or even just receives a signal that a domain controller is about to lose power—the data collection module automatically activates its emergency protocol.

Let's use a common collision scenario as an example. Within the EXD edge computing engine, vCompute, we deploy a sophisticated collision detection and identification model. This model uses a multi-signal fusion approach to maximize its coverage, accuracy, and reaction speed.

Specifically, it analyzes signals across three distinct phases:

• Pre-Collision: The perception system (millimeter-wave radar, ultrasonic sensors, object fusion) tracks the changing distance to obstacles.

• During Collision: High-frequency vehicle dynamics are monitored, including six-axis acceleration, yaw rate, pitch angle, driver inputs (braking, acceleration, steering torque), and vehicle state (gear, speed, master cylinder pressure).

• Post-Collision: The driver’s actions (e.g., stopping the car, activating hazard lights, locking the doors) are used as supplementary data to confirm the event.

By dynamically modeling these diverse signals, we can build a collision-severity model that accurately distinguishes between high-impact crashes, low-speed scrapes, and non-critical disturbances, effectively avoiding the false alarms common with traditional, single G-sensor trigger methods.

Emergency Storage Protocol

Once triggered, the EXD data collection module fundamentally changes its storage strategy, immediately taking the following steps:

1. It instantly flushes all data currently held in the memory cache to the disk, ensuring no cached data is lost.

2. It appends a Cyclic Redundancy Check (CRC) signature to the saved file, guaranteeing its integrity and making it tamper-proof.

3. It initiates a continuous write mechanism, saving any new, incremental data to the disk once every second until a pre-configured limit is reached or the vehicle loses power completely.

4. A few seconds later, it triggers a data-patching mechanism, which re-queries and uploads critical data fragments that may have been delayed by network latency during the chaotic event, ensuring no piece of the puzzle is missing.

This provides the ultimate safety net for reliable data forensics.

If the continuous write mechanism reaches its limit and the vehicle remains operational—meaning a worst-case scenario was narrowly avoided—the EXD module automatically reverts to its normal periodic storage mode. This routine mode uses an efficient columnar storage structure and data bucketing to balance performance with hardware longevity, ensuring data continuity during everyday driving.

The Complete Emergency Data Preservation Process

From the moment the vehicle starts, the EXD edge computing engine loads the event-detection models and begins real-time monitoring of all relevant signal streams. When an incident occurs, data is secured in four steps:

• Step 1: Immediate Save of Unsaved Data.

  The instant the algorithm running in vCompute detects a potential power-loss event, it commands the on-vehicle database, vData, to trigger an emergency flush of all data held in the memory cache. This prevents data loss just before the operating system shuts down. The saved file is signed with a CRC check to guarantee its integrity.

• Step 2: Back-Capture the "30-Second Black Box".

  Simultaneously, the collection module queries vData to extract and upload a complete, high-fidelity record of the 30 seconds prior to the event. This data, covering the vehicle's dynamics, perception status, and control decisions leading up to the incident, is the foundation for understanding the "causal chain" of the accident.

• Step 3: Initiate the Continuous Write Loop.

  To account for a potential delayed shutdown, the module then begins saving new data to the disk every single second. This continuous loop extends the data protection window for as long as possible, ensuring the vehicle's real-time state after the initial impact is also preserved and transmitted.

• Step 4: The Final Data Sweep.

  Five seconds after the initial trigger, the module performs one last query, requesting the data block from the five seconds immediately following the event. This clever step captures any critical data packets that may have been delayed by in-vehicle network latency during the crisis, guaranteeing a complete and coherent record.

This comprehensive process ensures that even if the vehicle loses all power within seconds, the system preserves the maximum amount of critical data from before, during, and after the event. This data is then uploaded to the cloud, providing an ironclad basis for liability assessment, accident reconstruction, and autonomous driving system optimization.

By leveraging EXD's edge computing and database capabilities for real-time risk detection, we can acquire millisecond-level data in the most extreme scenarios, helping automakers analyze the root cause of any incident. Furthermore, by building a tiered alert system, we can instantly notify after-sales and customer service departments, potentially cutting down emergency response times by precious minutes and creating a greater window for rescue.

As intelligent vehicles become increasingly software-defined, their perception, decision-making, and execution chains grow ever more complex. In the aftermath of an accident, the fine-grained operational details of these systems are the key to uncovering the truth, identifying problems, and advancing automotive safety.

At EXCEEDDATA, our goal is to ensure that in these critical moments, the data is never silent. It’s about protecting everyone in the car.