Fault Tolerance in Power Steering Systems (EPS): A Review for steer-by-wire systems

 Abstract

Among the X-by-Wire innovations, Steer-by-Wire (SbW) emerges as the game-changer poised to radically alter the car industry. An SbW system comprises electronic control units, sensors, and steering aid motors, offering the capability to replace a car's mechanical steering column links. However, before SbW systems can be widely adopted, several challenges need to be addressed. Two of the most critical ones are system reliability and fault tolerance. While Fault Detection and Isolation (FDI) has been extensively researched, fault diagnosis and fault tolerance in SbW systems have received less attention.

 1. Introduction

Most of our everyday mobility needs are met by automobiles. It is challenging to limit the frequency of fatal accidents due to the rising number of cars and the increasing need for mobility without adequate preventive measures. The greatest cause of death among those aged 15–29 is traffic accidents, which account for over 3,287 deaths per day globally, according to the Association for Safe International traffic Travel (ASIRT). Therefore, the most important thing for contemporary cars to consider while designing them is passenger safety. Factors such as driver error, car flaws, road conditions, environmental factors, etc., can all contribute to the occurrence of accidents. Fatal road accidents can occur as a result of these features' lack of directional control and the vehicle's possible lane departure.
Unwanted vehicle motion in response to a steering order occurs in real life due to unforeseen environmental inputs and disturbances such side-wind force, tire pressure loss, weather, and road conditions. When cornering, external disturbances have a greater impact since the vehicle's lateral dynamic condition determines whether it oversteers or understeers. The development of sophisticated steering control systems and research into the dynamic properties of vehicles are thus necessary to ensure the vehicles' directional stability in the presence of external disturbances.
One of the most important parts of a car is the steering system, which is responsible for directing the vehicle in the direction the driver specifies. The vehicle's handling and stability are then determined by it. By connecting the driver's input via the hand wheel to the vehicle's wheels via the steering column and gear arrangements, a standard automotive steering system ensures the vehicle remains stable in its intended direction. Hydraulic, electrohydraulic, and electric power assisted steering systems are just a few examples of the many actuators, sensors, and embedded systems that have become an integral part of modern steering systems, all with the goal of making vehicles more maneuverable.

 Everything from the steering wheel and column to the manual gearbox assembly, pitman arm, rack and pinion, steering linkages, and wheel spindle assemblies work together to direct a vehicle using human power. The drag link transfers the steering effort from the steering box to the wheel, which in turn transmits motion to the steering box. A tie-rod connects the two stub-axles. With the use of a steering box, you can reduce the gear ratio, allowing you to maneuver with minimal effort. Figure 1 shows the two kinds of steering mechanisms utilized in manual steering systems: worm and roller and rack and pinion. For parking and cornering, the driver will need to exert more steering effort. Figure 2 shows the hydraulic circuits—hydraulic power piston and control valve—used to create power assisted steering, which reduces the amount of effort required by the driver. An external source of power, provided by a hydraulic pump, is required by the power steering system. The pressure in the hydraulic fluid is used by the power piston to provide the forces that are needed to turn the wheels. When the steering is not applied, the control valve remains in its center position thanks to the central springs. In this position, the hydraulic pump returns the fluid to the reservoir tank. The control valve is pushed to the right side by the central spring as the steering wheel turns anticlockwise. This allows the hydraulic pump to function on the rack piston's right side. The driver's effort to turn the drop-arm clockwise is aided by the fluid. The amount of force exerted on the drop arm is proportionate to the steering effort.

Figure 1: Rack and Pinion Structure

Figure 2: Hydraulic Power Assisted Steering

Figure 3: Electric Power Assisted Steering

Figure 4: Electro-Hydraulic Power Assisted Steering

A hydraulic steering pump, which is connected to the belt drive, spins nonstop until the engine starts to turn over the steering fluid. Because of this, the engine's efficiency is diminished. An electrically operated pump can lessen the requirement for pumps, motors, and valves while also cutting down on friction losses. The ability to manage the power steering pump is the key differentiator between electro hydraulic systems and traditional hydraulic systems. Pump control is now accomplished by means of an electric motor rather than a belt drive, as illustrated in Figure 4. The steering mechanism is powered by pressurized oil and is activated when the driver applies the pedal to the wheel.

Researchers have developed a method called "Electric Power Steering" to enhance the steering system's performance, as illustrated in Figure 3. Instead of using hydraulic components, this steering system relies on an electric motor to provide direct assistance to the driver. Column, pinion, double-pinion, and rack electric power steering systems are all on the market.

 (a)       Rack and Pinion Structure

(b)        Hydraulic Power Assisted Steering

(c)        Electric Power Assisted Steering

(d)       Electro-Hydraulic Power Assisted Steering

Though there are many steering systems available, the flexibility of design is still a disadvantage in a conventional steering system. When an impact occurs, the damages could be fatal for the driver. The friction errors will also be high due to the mechanical connections between the steering wheel and the front wheels. To overcome the problems, the researchers have come up with the advanced steering technique called ‘Steer-by-wire’.

The technical advancements in automation industry have given many improved techniques in steering systems. In a steer-by-wire system, the mechanical connection between the hand wheel and the road wheels is replaced by an electric motor attached to the rack and pinion and an electronic control unit as shown in Fig5 .

Figure 5: Steer-By-Wire Architecture

The electronic control unit (ECU) in a steer-by-wire system receives inputs from torque and steering angle sensors. These signals are used to determine the wheel's reactive torque and angular position. The ECU then sends signals to the front wheel motor, which directly steers the front wheel. To simulate the sensation of steering, feedback on road tire interaction and vehicle dynamics is sent to the steering wheel motor. This allows the system to effortlessly achieve variable steering ratios. For example, a higher ratio at low speeds for better maneuverability and a lower ratio at high speeds for stability, without complex gear mechanisms. Integrating sensors and actuators into the steering system offers several advantages. These include active steering, variable steering ratios, and reduced steering effort.

2. Common Control Architectures for Steer-by-Wire Systems:

A number of control architectures have been suggested by different academics to address the needs of the steer-by-wire system. In order to keep the steer-by-wire system's stability constant, Segawa et al. (2001) suggested an automated steering control system (Fig. 6). Using the vehicle's control settings, researchers conducted experimental studies in the driving simulator and investigated parameter control optimization.

In  Peretti et al. ,(2005) presented magnetorheological fluids as a substitute for mechanical links that provide input to the driver. It was a handmade MR fluid that was semi-active and attached to the steering wheel; it served as a force feedback device and was inexpensive. 

Figure 6: Automatic steering control of steer-by-wire (Segawa et al., 2001)

By including integrated vehicle dynamics, Coudon et al. (2006) created a fresh steer-by-wire system reference model. A lateral force and front wheel kinematic model from a bicycle was used to fine-tune the model. A virtual force, obtained from the optimum feedback controller loop, was used to finish the procedure.

A steer-by-wire system's optimal feedback torque design was created by Gualino et al. (2006). It had been fine-tuned based on how the steering felt. System bandwidth, inertia effects, and the phase relationship characteristic were the primary areas of attention.

The torque feedback was created by Tahami et al., (2009) utilizing fractional order modeling of complicated dynamical systems based on nonlinearity. In order to simulate the vehicle, we used a single track model and computed the self-aligning moment for the feedback. The frequency response of the system was shown using the simulation results.

Figure 7 shows how Mehdizadeh et al. (2011) used the virtual vehicle idea to mimic the force feedback seen in traditional steering systems. Additionally, it was mentioned that the current approaches have an impact on a lanekeeping assistance controller's performance, which the new way suggests may be decreased. Also displayed and contrasted with the traditional steering system were the results of the simulations.

Figure 7:Virtual vehicle concept (Mehdizadeh et al., 2011)

Bajcinca et al. (2006) utilized the Stochastic Gauss-Markov method with a Kalman filter to estimate steering wheel friction forces. The model considered two scenarios: rectangular input to the steering wheel and rectangular input with disturbance to the front wheels. A two-degrees-of-freedom reference vehicle incorporating admittance control and the steer-by-wire system's nonlinear vehicle dynamics was employed for testing.

Fahami et al. (2015) developed a current-based control approach for force feedback. Their system incorporated the tire's compensating torque using a linear quadratic regulator, leading to improved performance..

3. Techniques for Torque Feedback Estimation in Steering Feel Generation:

In order to generate steering feel and estimate torque feedback, many variables pertaining to the vehicle's dynamic states, road-tire interactions, road conditions, etc. must be considered. To provide the torque feedback for the driver, a complete model is needed. Research on torque feedback estimates and control in steer-by-wire systems has made use of a variety of methods, which are detailed in the sections that follow.

TRM Approach

In order to provide the sensation of steering in the steering wheel motor, Oh et al. (2004) suggested a torque map method that relies on the control parameters of vehicle speed and steering wheel angle (Fig. 8). Using the map, we were able to determine the optimal torque, and the driver has complete control over the sensation.

 

Figure 8 Torque map approach (Oh et al., 2004)

TORQUE SENSOR METHOD

Torque sensors are a widely used tool for measuring and providing feedback that allows drivers to experience a more customizable driving feel. To enhance steering feedback specifically, Kim et al. (2008) proposed(figure 9) incorporating a torque sensor into the steering wheel controller.

Figure 9: Steer-by-wire using torque sensor (Kim et al, 2008)

The loop shaping approach for reactive torque creation was suggested by Sun et al. (2006) and is displayed in Figure 10. Strong control methods for improving the driver's perception of torque. The ripple reduction and tracking were both handled by the H_∞ controller.

 

Figure 10: Loop shaping technique in steer-by-wire (Sun et al, 2006)

The torque sensor system suggested by Chen et al. (2013), shown in Figure 11, uses a H∞  controller to provide feedback to the driver. Time and frequency domains were also analyzed.

.

Figure 11: H_∞ in steer-by-wire (Chen et al., 2013)

The steering wheel's torque feedback system has made use of a variety of sensors, including strain gauges and torsion bars. These torque sensors are useful, but they're expensive and need to be calibrated often. The development of a torque feedback estimate technique based on the real vehicle's inexpensive sensors is, therefore, essential..

 MODEL BASED METHODS

Researchers have suggested model-based approaches that take into account various vehicle dynamic factors of the steer-by-wire system in order to provide the driver varied input for changing road conditions. Using a variety of vehicle dynamic characteristics, including steering wheel angle and yaw rate, Shengbing et al. (2007) introduced a model for estimating steering feedback. Figure 12 shows the results of an adaptive controller for vehicle steering that Yamaguchi et al. (2009) created. When the front tire stiffness changed, the control method was used. For the driver's perception of the road, the estimated self-aligning torque is supplied back.

Figure 12: Hardware setup of steer-by-wire vehicle (Yamaguchi et al., 2009)

To put the idea of steer-by-wire force feedback into practice, Setlur et al. (2003) developed a new model for autonomous cars using VR technology. To make sure the vehicle reacts to the driver's commands and gives them enough feedback, the non-linear tracking haptic controller is responsible. The simulation results showed that the proposed method had a good chance of giving the driver more feedback and control.

A new control mechanism for rack-actuated steer-by-wire systems was introduced by Jin Park et al. (2005) to remedy the drawbacks of conventional steering systems. In their hardware-in-the-loop simulation, they included the steering wheel and front wheel motor as part of their bond graph model for monitoring and feedback. A PI controller was installed in the steering wheel model, and the experimental work made use of a DC brushed motor to enhance the driver's feedback.

 4. Benefits of CLOUD AND SOLUTIONS

The problem arises when something isn't working properly, things aren't going according to plan, or bad actors are involved, according to Rafael et al. (2006). According to Rohan et al. (2012), system tolerance is defined as the presence of mechanisms to manage such errors in order to maintain the system's ongoing operation.

Figure 13: Fault Tolerance Taxonomy

As illustrated in Figure 13, a taxonomy for fault tolerance exists, encompassing mechanisms for fault management, fault handling prototypes, and failure types. The significant heterogeneity inherent in hybrid cloud environments appears to be the primary culprit behind the decline in reliability and dependability trends. The possibility of subtle error propagation between nodes exists, with the entire system potentially cascading into disrepair due to hypervisor malfunction and disparity among nodes (Cheng et al., 2012). This suggests, at the very least, a decrease in the reliability of modern hybrid clouds. Furthermore, as the number of unforeseen Byzantine failures increases, it becomes increasingly difficult to predict failure behaviors at the node or transient level within the cloud. Intervention plans have yet to yield the desired outcomes.

Cloud Faults:

Common Causes of Faults in Cloud Computing

Research by Schroeder, Gibson (2010), and Lakshmi et al. (2014) identified common causes of faults in cloud computing, including:

* Computer hardware failure or malfunction

* Software errors

* Human errors

* Network issues (diagnosis and resolution)

* Security breaches

* Virtual component failure

* Virtual connection failure

Types of Cloud Faults

Cloud processing nodes can experience temporary faults in communication channels between themselves or with other nodes. Three main types of cloud faults exist, as identified by Dominic et al. (2005) and Long et al. (2010):

·         Inter-node faults: According to Ifeanyi et al. (2013), an inter-node is a node with a TCP/IP protocol stack connection to another cloud node. Transient faults with inter-nodes are typically one-time occurrences that dissipate due to the inherent resilience of these connections. Troubleshooting is usually not required.

·         Intra-node faults: Sunay et al. (2009) define intra-nodes as nodes within the same system linked by temporary connections. These nodes are highly virtualized and do not adhere to the TCP/IP stack. Transient faults in intra-nodes are more persistent, often requiring manual troubleshooting.

·         Byzantine faults:  According to Ifeanyi et al. (2013), intermittent faults can disrupt system and device operation at irregular intervals.  Miguel Barbara (1999) describes intra-node intermittent faults as more complex than inter-node faults, potentially evolving into byzantine faults. Martin (2006) highlights the difficulty of detecting these faults because they can produce seemingly valid outputs regardless of the error.  While incorporating features like alarms and error-avoidance mechanisms into the system is possible (Widodo et al., 2017), the lack of detection systems hinders training virtual machines to handle such errors.

·         Permanent Faults in Virtual Nodes:

Most cloud failures occur within intra-nodes and are considered permanent until the malfunctioning virtual component is addressed. This differs from the standard repair or replacement procedures used for inter-node faults. A study by Chunye et al. (2010) found that virtual components often collaborate to complete cloud tasks. Even temporary or permanent outages in a single component can disrupt the entire operation.

·         Fail-Silent vs. Byzantine Failures

These errors can result in either fail-silent or Byzantine failures. Fail-silent failures are predictable and easily detectable because malfunctioning components either cease operation entirely (no output) or produce substandard, readily observable output (Ravi Incenzo, 2013).

 

Metrics for Fault Tolerance

Cloud computing's current approach to fault tolerance considers several factors:

Ø  Reaction time: How quickly the system can identify and respond to a fault.

Ø  Availability: The uptime or accessibility of the system.

Ø  Performance: The speed and efficiency of system operations.

Ø  Scalability: The ability to handle increased workloads by adding resources.

Ø  Security: Protection against unauthorized access and data breaches.

Ø  Usability: The ease with which users can interact with the system.

Ø  Dependability: The reliability of the system to function as expected.

Ø  Overhead: The additional resources consumed by fault tolerance mechanisms.

5. Byzantine Faults in Distributed Systems

Distributed systems can sometimes encounter Byzantine faults. These faults cause the system to produce both correct and incorrect outputs, leading to failures. When a client interacts with a system experiencing a Byzantine fault, the response may be erratic. This can include incorrect execution of instructions or even system crashes. Byzantine faults are not only confusing but also misleading. They can cause even functioning components to malfunction, making it difficult to pinpoint the root cause of the problem.

Byzantine Fault Tolerance

Fortunately, distributed systems can be built to tolerate Byzantine faults. This is achieved through a technique called replication, where all services and their copies are guaranteed to agree on the system's state at any given time. Replication safeguards the system against Byzantine faults by ensuring that correct data reaches all components.

The Overarching Issue with Byzantine Systems

An enormous distributed system can malfunction in several ways. One common issue is omission failures, where a node fails to respond to or receive a request. Another type of failure occurs when the data sent is inaccurate, leading to corrupted local state or an erroneous response.

Failure Detectors

In Byzantine systems, failure detectors are used to identify malfunctioning nodes. These detectors label nodes as either trustworthy or untrustworthy based on their behavior. A dependable failure detector produces accurate and timely results. Detectors that are slow or provide inaccurate information are considered unreliable. This latter group encompasses the majority of failed attempts.

Figure 14: Proposed Byzantine Fault Detection Zone

 

Despite the possibility of random process failures, a failure detector should still ensure several key qualities (as depicted in Figure 14). Here, we will examine the metrics that define a failure detector's Quality of Service (QoS).

Comprehensiveness: When a failure detector definitively identifies a process failure, it indicates completeness. Other unsuccessful operations serve as evidence of a specific process malfunction.

Precision: A failed process must be accurately identified as such. It is impossible to create a perfect failure detector for a real-world network. While they may be incomplete or probabilistic, real detectors strive for 100% accuracy.

Timeliness: The time it takes to detect a failure should be minimal.

Scalability: The overall network load should be minimal, and the strain on individual processes should be modest and evenly distributed.

Detection Time (TD): This refers to the time elapsed between a process p1's crash and another process p2's suspicion of p1's permanent failure.

6. FAULTS DETECTION   AND TOLERANCE

Jun et al. (2010) introduced methods for reducing page faults, categorized into read fault reduction and write fault prediction. Further observations could have assessed their ability to handle Byzantine faults.

Dominic et al. (2005) listed three objectives for fault tolerance. The first is to increase fault tolerance without altering current system operation. The second is to run non-fault-tolerant software as quickly as possible. Finally, the third objective is to reduce the space and time overhead required to identify and recover from problems. ExtraVirt (Dominic et al., 2005) achieved these goals by leveraging virtual machine technology to share memory and input/output devices between replicas.

Challenges of Byzantine Faults

Lakshmi Yumnam (2014) highlighted the challenge of maintaining performance levels required to meet Quality of Service (QoS) agreements. The most difficult aspect of Byzantine faults is the attacker's ability to disguise a breach as a Byzantine fault, often leading to significant harm.

As Kevin et al. (2003) pointed out, Byzantine faults can surreptitiously infiltrate the Cloud environment without detection. This has the potential to rapidly infect additional virtual machines. The system continues to function despite the generated defects. While other fault types might be identified, Byzantine faults remain elusive despite their destructive potential. According to Hiep et al. (2011), Byzantine faults can be used as a transmission mechanism to cause complete cloud failures across virtual machines, networks, and applications. Cloud services offer a pay-as-you-go model. The biggest concern for both Cloud Service Providers (CSPs) and their clients is the issue of increased costs resulting from errors.

Importance of Byzantine Fault Detection

Detecting Byzantine faults is crucial. Even a single instance should be identified promptly to prevent the error from compromising the checkpoint and spreading further. Proactive detection is essential to catch Byzantine faults early and prevent them from causing damage or propagating across the cloud system. Even with well-defined proactive detection methods offering 99% discovery confidence, there is still a 1% chance that a Byzantine fault might infiltrate the cloud system. The small Byzantine faults that are produced might trigger the checkpoint reached for the prior interval fault, making fault tolerance in such scenarios much more challenging. Consequently, the proposed system strives to demonstrate the accuracy of Byzantine fault detection techniques.

Checkpointing for Fault Tolerance

The literature review by Zhou et al. (2017) identifies checkpointing as one of the most popular fault tolerance mechanisms. Checkpointing approaches can be rapidly deployed when a virtual machine (VM) or group of VMs experiences a failure because they record the state of each VM as an image file. Manav et al. (2011) discuss maintaining consistent checkpointing in an error-prone environment when a failure is predicted. This requires rigorous checkpointing at minimal intervals. Subba and Ramesh (2014) note that as intervals shrink, space and time consumption decrease. However, for large data analytics applications, checkpointing can be broken down into individual jobs (Dursun, 2013.

Cloud Computing and Large-Scale Data Processing

Cloud computing systems are frequently employed for large-scale data processing, particularly big data analytics, due to the immense processing requirements that on-premise infrastructure cannot handle.  (Ibrahim et al., 2015)

 

In these scenarios, automated allocation of input data into smaller, manageable chunks, often called tasks, is crucial (Ziqian et al., 2015). These tasks can be further categorized and optimized for processing by virtual machines (VMs) (Cheng et al., 2012).

However, it's important to note that tasks don't always represent entire occupations; they can be subdivisions of larger jobs. Dividing work into smaller tasks allows for efficient handling by VMs, which can only execute applications that have been subdivided through an n-level process. (Cheng et al., 2012)

Virtualization and Fault Tolerance in Cloud Systems

In a highly virtualized cloud environment, multiple VMs can share a single physical host and communicate with each other via bridges. This allows for consolidation of multiple users and applications within a single system. Importantly, processes running on one VM are isolated from processes on other VMs, meaning a failure in one VM doesn't affect the operation of others.  Checkpointing allows processes to be migrated to other VMs in the event of a failure. However, due to the complexity of detecting certain errors (Byzantine failures), checkpointing is often abandoned in favor of migrating the entire job, rather than individual tasks, when an error is identified in a single VM. (Yilei et al., 2011)

Byzantine Fault Tolerance (BFT) in Cloud Computing

Yilei et al. (2011) proposed a BFT Cloud architecture that can function in a voluntary resource cloud by selecting nodes based on their quality-of-service (QoS) performance. While BFT improves QoS, it can introduce subtle output errors, potentially leading to an endless loop in the system. Pedro et al. (2011) introduced a MapReduce approach and prototype that can handle Byzantine failures. This method can be further improved for larger cloud environments.

7..Fault detection and isolation (FDI)

 Fault Detection and Injection (FDI) in Fault-Tolerant Control Systems

 Fault detection and injection (FDI) is a crucial component of fault-tolerant control systems. It identifies malfunctions within the system and relays this information to the controller. The controller can then take corrective actions to mitigate or fix the problem, ensuring the overall performance of the system remains intact. FDI acts as a watchful guardian, constantly monitoring the system for anomalies. Upon detecting an issue, it not only identifies the problem type but also pinpoints its location. There are two primary approaches to address FDI for Steer-by-Wire (SbW) systems: hardware redundancy-based and analytical redundancy-based. We will delve deeper into these methods below.

Hardware Redundancy

Hardware redundancy is a fundamental requirement for most fault-tolerant control systems because it offers the simplest way to achieve safety and reliability targets. This approach involves adding additional modules, often placed in parallel with a specific module. By comparing the duplicated output signals, hardware redundancy facilitates fault diagnosis. The core concept behind this method lies in the ability to distinguish between faulty and healthy outputs (Bertacchini et al.,   2005). In most cases, a majority voting technique is employed to determine the timing and probable location of the problem (Anwar & Chen, 2007). Once a fault is detected, the signal from the malfunctioning component is disabled. However, there are limitations to this approach. For instance, a faulty voter itself can lead to erroneous majority results, disrupting system functionality..

Mechanical Support

A mechanical backup mechanism exists to re-engage the steering wheel and actuator in case the steering actuator fails. This essentially reverts the vehicle's steering to a traditional mechanical system. However, this backup system may not activate promptly in situations where steering failure occurs, potentially leading to collisions.

Steering Motor Hardware Redundancy

The SbW system utilizes two motors: a steering actuator motor and a feedback motor , which are responsible for generating feedback torque and steering torque, respectively. Due to the critical role these motors play, numerous researchers and automotive companies have implemented redundancy measures to mitigate the impact of motor failure in SbW systems. Zong et al, (2012) depict an SbW system design that incorporates two steering motors. This configuration enhances fault tolerance by providing actuator redundancy. In the event of a failure in one actuator, the other can continue to function independently, maintaining steering capability.

It's important to note that these studies utilize the FlexRay bus instead of the CAN bus employed by Zong et al. (2012). The FlexRay bus offers superior handling of temporal communication delays and faults. However, none of these aforementioned studies address fault diagnosis, leaving a gap in determining when to activate fault-tolerant mechanisms.  Implementing a dual-motor design introduces an additional challenge for the control system. The system must not only manage the movements of both motors concurrently but also ensure they don't interfere with each other during normal operation.

Mei et al. (2007) studies the three-level hardware redundancy of a steering actuator that uses a permanent magnetic brushless DC motor. One downside of DC motors is their inadequate availability, but they have two advantages—less noise and longer motor life. An SPM synchronous motor provides torque feedback, while a permanent magnet synchronous (PMS) motor controls the steering in the system designed by Benedetti et al. (2005). Permanent magnet synchronous motors (PMS motors) are easily recognizable by their tiny package size, high efficiency, and fault tolerance. Electronic speed drives for DC and induction motors, as well as various hardware configurations, are reviewed in the 2008 work of Campos-Delgado, Espinoza-Trejo, and Palacios.

 
Hardware redundancy for feedback actuator

When it comes to simulating the reactive forces acting on the wheel, the force feed-back motor outperforms the steering actuator motor. The feedback motor control technique is implemented by Bertacchini et al. (2005) using hybrid redundancy as a hot stand-by redundancy in the case that the triple modular redundancy architecture fails.

 
As shown by Krautstrunk and Mutschler (2000), a standard three-phase permanent magnet synchronous motor (PMSM) may be used as a fault-tolerant force feedback motor. For the sake of safety and dependability, the configuration is converted from three-phases to two-phases operation in event of a single phase failure.

Hardware redundancy for sensors

To further ensure that the electronic control unit of a SbW system can dependably receive the signals needed to evaluate the vehicle's status and respond appropriately, sensor fault tolerant control is included. In order to determine whether a failure has happened, hardware redundancy compares data from several sensors and uses a voting mechanism.

Using duplicated sensors, he et al. (2015) describe the hardware designs for SbW systems. When one operational sensor fails, a backup sensor will step in to keep the SbW system's electronic control unit apprised of the vehicle's condition and allow it to respond quickly enough.

Hardware redundancy for electronic control unit (ECU)

Sensors and actuators in SbW systems need to use the same redundancy methods as the ECU to guarantee safe operation. If a single redundant ECU fails, the system will switch to the other one's mode of operation. In this case, it is crucial to have reliable failure detection that is quick. We shall connect the redundant ECUs in parallel so they may run simultaneously if this is not practicable..  This may reduce the time it takes to switch over in the event that the main unit fails (Pimentel, 2004).

Hardware redundancy for communication protocol

Several safety functions, including failure detection, reconfiguration, and recovery strategies, are the key aspects of Pimentel (2004). The software architecture and hardware redundancy design achieve these goals by making use of several duplicated components for sensors, CAN buses, controllers, and actuators. Meeting the fault tolerance, recoverability, and fail-safety criteria for SbW systems may be achieved with the help of enormous replication and safety-critical software design.

Analytical redundancy

Reducing the total cost of manufacturing while maintaining dependability is the goal of analytical redundancy, which in turn aims to make SbW system manufacture economical.
Fig. 15 shows a high-level architecture of an FDIR system that relies on analytical redundancy for fault detection, isolation, and reconfiguration. To determine the estimate of target variables in analytical redundancy-based FDIR, the mathematical SbW system model is used in its analytical version.  

Figure 15: FDIR's analytical redundancy and hardware .

Residual-based FDI

Under typical circumstances, the residual—the difference between the measured and calculated values of a variable—has a mean of zero. To find and fix problems before they negatively impact the vehicle's steering and handling, residuals should be fault-sensitive. To further aid in fault isolation, each residual should be noise-and uncertainty-insensitive while being very sensitive to the target fault (Fig. 16).

Figure 16: Residual-based FDI.

A common method for collecting diagnostic residuals in fly-by-wire systems of airplanes is to use three sets of redundant sensors; other steer-by-wire studies have proposed the same idea (Führer & Schedl, 1999). Analytical redundancy makes it possible to produce residuals even when physical sensors are not available.

 In most cases, each residual might be impacted by a myriad of possible fault scenarios. For example, if you possess the residuals of the predicted electrical resistance or motor constant of the steering motors, you may use them to detect a failure in the motor current sensor (Li, Zhao, He, and Lu, 2019). A steering controller failure and a dead battery may be detected using the tracking inaccuracy of the anticipated motor current. Additionally, you may use the estimated front wheel angle residuals to identify a yaw rate sensor failure, a front wheel angle sensor failure, or a battery failure. An adequate amount of residuals is necessary to enhance the defect detection rate and fault isolation level.

 Two of the predicted signals—the electrical resistance and the motor current—can only be extracted from readings made by sensors that are already part of a SbW system. Estimated signals, such as front wheel angle, must be formed using state estimation methods.

Unknown input observer (UIO)

 The primary goal of UIO is to provide a collection of decoupled residuals that are fault-sensitive and to minimize or eliminate the impact of unknown disturbances on the operation. Residuals that are both noise-resistant and fault-specific are produced by using the fault isolation banks of the UIOs described in Dos Santos et al. (2016). Because of this, problems with the in-wheel motor or steering of SbW systems may be located and fixed very precisely. The UIO's main limitation is that it can't detect and isolate several errors at once.

Sliding Mode Observer

The sliding mode observer is characterized by its ability to estimate state variables with minimal influence from external disturbances or uncertainties. Due to the time-varying and non-linear dynamics inherent in both real SbW systems and the vehicle itself, a sliding mode observer is developed based on the non-linear SbW model (Anwar & Niu, 2010). This approach has been used to detect multiplicative faults within the SbW system . A key distinction between sliding mode observers and other types is their fixed-time convergence property, ensuring that the estimation error for all estimated states reaches zero within a specific timeframe.

Kalman Filter and Recursive Least-Squares Estimator

Several studies, including those by Xu et al. (2018) employ the Kalman filter and the recursive least-square estimator to generate residuals capable of differentiating between various fault conditions, encompassing actuator, sensor, controller, and battery failures. Notably, all SbW system residuals derived from these methods rely on measurements from standard steering system sensors, making them susceptible to a wide range of potential fault scenarios.

Parity Space Method

Rearranging the SbW system model structure using observed front wheel angle and known motor current data yields residual signals; this is the basic notion behind the parity space technique. Under ideal state operating circumstances, a non-zero residual or parity equation value indicates a malfunction, while a zero value indicates that all components are working flawlessly. It is possible to use this method to diagnose faults in sensors and actuators; moreover, it does not need fault knowledge and performs well in simulations (Moon et al., 2005). As a result of measurement noise, inaccurate models, and significant errors in sensors and actuators, the residuals are never zero in practice. Concerning uncertainties in multiplicative parametric errors and measurement noise, the parity space approach has to be strengthened.

Hidden Markov Model (HMM)

In order to estimate the exact value of the steering wheel angle and the steering angle velocity at the next time step, He et al. (2010a) employ HMM to determine the current driving state of the vehicle based on sensor data. Data from sensors and predictions made by the driving state forecasting controller are used in difference-value computations to produce the residual.

Fault detection:. During fault detection, the components are examined to ascertain the presence or absence of a defect. One easy way to find anything is to compare the residual with a fixed threshold. When the residual goes beyond the cutoff, it is said that there is a defect. An extremely difficult issue is the determination of a threshold. A low threshold could increase the false alarm rate, whereas a high one might cause non-detection. When an alarm goes off when there are really no problems, it is called a false alarm.

8. Conclusion

In this paper, we analyze and discuss a variety of fault-resistant control features for SbW systems. Researchers have explored numerous fault-tolerant algorithms and fault detection and isolation (FDI) techniques to address the high cost, safety-performance, and fault tolerance requirements of SbW systems. While these techniques have shown promise, several design issues remain before this field can fully mature. One critical challenge concerns the accuracy of FDI mechanisms within the FTCS architecture. The proper operation of an FTCS heavily relies on accurate fault diagnosis. However, model-based FDI techniques assume a perfect mathematical representation of the SbW system, which is impossible to achieve.. Modeling uncertainties and noise can significantly impact diagnostic performance, including accuracy, speed, and fault isolation time. To avoid false alarms, effectively isolating the effects of disturbances from the residual signal is crucial. Therefore, robust FDI techniques that can handle system failures caused by disturbances and modeling inaccuracies are an essential area for further research.Time delays are another significant source of instability and performance degradation in SbW systems.. Furthermore, if the delay exceeds the system's maximum permissible response time, vehicle safety becomes compromised . Interestingly, limited research explores the FDI of SbW systems affected by time-varying delays. Additionally, the time between fault occurrence and activation of the fault-tolerant controller is critical for ensuring safe operation. A lengthy FDI process can jeopardize the SbW system's integrity and degrade steering performance.

 

 

References

Anwar, S., & Chen, L. (2007). An analytical redundancy-based fault detection and isolation algorithm for a road-wheel control subsystem in a steer-by-wire system. IEEE Transactions on Vehicular Technology, 56(5), 2859–2869.

Anwar, S., & Niu, W. (2010). Analytical redundancy based predictive fault tolerant control of a steer-by-wire system using nonlinear observer. In Industrial technology (ICIT), 2010 IEEE international conference on (pp. 477–482). IEEE

Bajcinca, N., C. Nuthong and F. Svaricek (2006). Road feedback estimation for steerby-wire control. Proceedings of the IEEE International Conference on Control Applications, Munich, pp. 1288-1293, Germany

Benedetti, A., Bianchi, N., Bolognani, S., Pre, M., Molari, P., Morelli, P., et al. (2005). Pm motor drives for steer-by-wire applications. In Industry applications conference, 2005. Fourtieth ias annual meeting. conference record of the 2005: vol. 4 (pp. 2857–2864). IEEE.

Bertacchini, A., Pavan, P., Tamagnini, L., & Fergnani, L. (2005). Control of brushless motor with hybrid redundancy for force feedback in steer-by-wire applications. In Industrial electronics society, 2005. iecon 2005. 31st annual conference of IEEE (pp. 6–pp). IEEE

brahim Abaker Targio Hashem, IbrarYaqoob, NorBadrulAnuar, Salimah Mokhtar, AbdullahGani Samee Ullah Khan 2015, „The rise of “big data” on cloud computing: Review and open research issues‟, Elsevier Information Systems, Vol. 47, pp. 98-115

Chen, L., M. Bian, Y. Luo and K. Li (2013). Maximum tire road friction estimation based on modified dugoff tire model. International Conference on Mechanical and Automation Engineering, Jiujang, pp. 56-61, China.

Cheng Tan, Yubin Xia, Haibo Chen Binyu Zang 2012, „TinyChecker: Transparent Protection of VMs against Hypervisor Failures with Nested Virtualization‟, IEEE/IFIP 42nd International Conference on Dependable Systems and Networks Workshops (DSN-W

Cheng Tan, Yubin Xia, Haibo Chen Binyu Zang 2012, „TinyChecker: Transparent Protection of VMs against Hypervisor Failures with Nested Virtualization‟, IEEE/IFIP 42nd International Conference on Dependable Systems and Networks Workshops (DSN-W)

Chunye Gong, Jie Liu, Qiang Zhang, Haitao Chen Zhenghu Gong 2010, „The Characteristics of Cloud Computing‟, 39th IEEE International Conference on Parallel Processing Workshops, pp. 275- 279.

Coudon, J., C.C.D. Wit and X. Claeys (2006). A new reference model for steer-bywire applications with embedded vehicle dynamics. Proceedings American Control Conference, Minneapolis, pp. 3990-3995, USA.

Dominic Lucchetti, Steven K. Reinhardt Peter M. Chen 2005, „ExtraVirt: detecting and recovering from transient processor faults‟, Proceedings of the 20th ACM symposium on Operating systems principles, pp. 1-8.

Dominic Lucchetti, Steven K. Reinhardt Peter M. Chen 2005, „ExtraVirt: detecting and recovering from transient processor faults‟, Proceedings of the 20th ACM symposium on Operating systems principles, pp. 1-8

dos Santos, B., Lopes, A., & Araújo, R. E. (2016). Fault detection scheme for a road vehicle with four independent single-wheel electric motors and steer-by-wire system. In Advanced Vehicle Control (pp. 417-422). CRC Press.

Fahami, S.M.H., H. Zamzuri and S.A. Mazlan (2015). Development of estimation force feedback torque control algorithm for driver steering feel in vehicle steer-bywire system: hardware in loop, International Journal of Vehicular Technology, Vol. 2015, No. 1, pp. 1-17

Führer, T., & Schedl, A. (1999). The steer-by-wire prototype implementation: Realizing time triggered system design, fail silence behavior and active replication with fault-tolerance support. SAE transactions, 646-653.

Gualino, D. and I.J. Adounkpe (2006). Force feedback system design for the steer by wire: optimization and performance evaluation. IEEE International Transportation Systems Conference, Toronto, pp. 181-187, Canada.

Haluk Demirkan Dursun Delen 2013, „Leveraging the capabilities of service-oriented decision support systems: Putting analytics and big data in cloud‟, Elsevier Decision Support Systems, vol.55, pp. 412–421.

He, L., Chen, G.Y. & Zheng, H.Y. Fault tolerant control method of dual steering actuator motors for steer-by-wire system. Int.J Automot. Technol. 16, 977–987 (2015). https://doi.org/10.1007/s12239-015-0100-8

Hiep Nguyen, Yongmin Tan Xiaohui Gu 2011, „PAL: Propagationaware Anomaly Localization for Cloud Hosted Distributed Applications‟, ACM Proceeding Managing Large-scale Systems via the Analysis of System Logs and the Application of Machine Learning Techniques, No. 1.

Ifeanyi P. Egwutuoha, David Levy, Bran Selic Shiping Chen 2013, „A survey of fault tolerance mechanisms and check-point/restart implementations for high performance computing systems‟, Springer The Journal of Supercomputing, Vol.65, pp. 1302–1326

Jun Zhu, Wei Dong, Zhefu Jiang, Xiaogang Shi, Zhen Xiao Xiaoming Li 2010, „Improving the performance of hypervisor-based fault tolerance‟, IEEE International Symposium on Parallel Distributed Processing, pp. 44-51

Kevin Driscoll, Brendan Hall, Håkan Sivencrona Phil Zumsteg 2003, „Byzantine Fault Tolerance from Theory to Reality‟, Springer International Conference on Computer Safety, Reliability, and Security, pp. 235-248

Kim, I and J. Lee (2008). A study of steering system model for steering feel improvement of steer-by-wire system, Citeseer, Vol. 23, No. 1, pp.1-6.

Klehmet, U., Herpel, T., Hielscher, K. S., & German, R. (2008, March). Delay bounds for CAN communication in automotive applications. In 14th GI/ITG Conference-Measurement, Modelling and Evalutation of Computer and Communication Systems (pp. 1-15). VDE.

Lakshmi Prasad Saikia Yumnam Langlen Devi 2014, „Fault Tolerance techniques and algorithms in cloud computing‟, International Journal of Computer Science Communication Networks, Vol. 4, No. 1

Lakshmi Prasad Saikia Yumnam Langlen Devi 2014, „Fault Tolerance techniques and algorithms in cloud computing‟, International Journal of Computer Science Communication Networks, Vol. 4, No. 1.

Li, H., Wang, Y., Yao, D., & Lu, R. (2018). A sliding mode approach to stabilization of nonlinear Markovian jump singularly perturbed systems. Automatica, 97, 404–413.

Li, H., Zhang, Z., Yan, H., & Xie, X. (2018). Adaptive event-triggered fuzzy control for uncertain active suspension systems. IEEE Transactions on Cybernetics. doi: 10.1109/TCYB.2018.2864776.

Li, H., Zhao, S., He, W., & Lu, R. (2019). Adaptive finite-time tracking control of full states constrained nonlinear systems with dead-zone. Automatica, 100, 99– 107

Long Wang, Zbigniew Kalbarczyk, Ravishankar K. Iyer Arun Iyengar 2010, „Check-pointing Virtual Machines Against Transient Errors‟, IEEE 16th International On-Line Testing Symposium (IOLTS), July 2010.

Manav Vasavada, Frank Mueller, Paul H. Hargrove Eric Roman 2011, „Comparing different approaches for Incremental Check-pointing: The Showdown‟, Proceedings of the Linux Symposium

Martin J.P. L. Alvisi 2006, „Fast Byzantine Consensus‟, „IEEE Transactions on Dependable and Secure Computing‟, Vol. 3, Iss. 3

Mehdizadeh, E., M. Kabganian and R. Kazemi (2011). A new force feedback for steer by wire vehicles via virtual vehicle concept. 50th IEEE Conference on Decision and Control and European Control Conference, Florida, pp. 2281-2286, USA.

Mei, T., Shafik, M., Lewis, R., Walilay, H., Whitley, M., & Baker, D. (2007). Fault tolerant actuation for steer-by-wire applications. In Automotive electronics, 2007 3rd institution of engineering and technology conference on (pp. 1–8). IET

Miguel Castro Barbara Liskov 1999, „Practical Byzantine Fault Tolerance‟, USENIX Proceedings of the Third Symposium on Operating Systems Design and Implementation, pp. 173–186

MOON, K. H., KANG, J. S., LEE, T. J., & JUNG, J. H. (2005). Two-stage revision surgery of the infected hip bipolar hemiarthroplasty using the PROSTALAC. The Journal of the Korean Orthopaedic Association, 442-448.

Oh, S.W., H.C. Chae., S.C. Yun and C.S. Han (2004). The design of a controller for the steer-by-wire system, JSME International Journal, Vol. 47, No. 3, pp. 896-907.

Park, T.J., C.S. Han and S.H. Lee (2005). Development of electronic control unit for the rack actuating steer-by-wire using the hardware-in-the-loop simulation system, Mechatronics, Vol. 15, No. 8, pp. 899-918

Peretti, L. and M. Zigliotto (2005). A force feedback system for steer by wire applications based on low cost mr fluids: design hints. The 3rd IET International Conference on Power Electronics, Machine and Drives, Dublin, pp. 469-473, Ireland

Pimentel, J. R. (2004). An architecture for a safety-critical steer-by-wire system.

Rafael R. Obelheiro, Alysson Neves Bessani, Lau Cheuk Lung Miguel Correia 2006, „How Practical are Intrusion-Tolerant Distributed Systems?‟, Department of Informatics, University of Lisbon

Ramesh Babu D. V. Subba Rao 2014, „Automatic check-point based fault tolerance in computational grid‟, IEEE Conference COMMANTEL

Ravi Jhawar Vincenzo Piuri 2013, „Fault Tolerance and Resilience in Cloud Computing Environments‟, 2nd Edition Morgan Kauffmann, Computer and Information Security Handbook.

Rohan Garg, Komal Sodha Gene Cooperman 2012, „A Generic Checkpoint-Restart Mechanism for Virtual Machines‟, arXiv:1212.1787

Schroeder, B. Gibson, G.A 2010, „A Large-Scale Study of Failures in High-Performance Computing Systems‟, IEEE Transactions on Dependable and Secure Computing, Vol. 7, Iss. 4.

Segawa, M., S. Nakano., O. Nishihara., and H. Kumamoto (2001). Vehicle stability control strategy for steer-by-wire system, JSAE Review, Vol. 22, pp. 383-388.

Setlur, P., J. Wagner., D. Dawson and L. Powers (2003). A hardware in the loop and virtual reality test environment for steer by wire system evaluations. Proceedings of the American Control Conference, Denver, pp. 2584-2589, Colorado

Shengbing, Yang & Chunan, Den & Ji, Xuewu & Kuiyuan, Chen. (2007). Research on Road Feeling Control Strategy of Steer-by-Wire. 10.4271/2007-01-3652.

Sun, X. D., P. G. Scotson, A. Luengen and H. D. Heitzer (2006). Development of 2 degrees of freedom steer by wire hand wheel actuator control. Proceedings of the 2006 IEEE International Conference on Control Applications, Munich, pp. 2814- 2819, Germany.

Sunay Tripathi, Nicolas Droux, Thirumalai Srinivasan Kais Belgaied 2009, „Crossbow: From Hardware Virtualized NICs to Virtualized Networks‟, Proceedings of the 1st ACM workshop on Virtualized infrastructure systems and architectures, pp. 53-62’

Tahami, F. and H. Afshang (2009). A fractional order model for steer by wire systems. 35th Annual Conference on IEEE Industrial Electronics, Porto, pp. 4161- 4166, Portugal.

Widodo N.S, Ryan Lim, Hyotaek Atiquzzaman, Mohammed. (2017). A new content-defined chunking algorithm for data deduplication in cloud storage. Future Generation Computing System. V: 71, PP. 1-234

Xu, F. X., Liu, X. H., Chen, W., Zhou, C., & Cao, B. W. (2018). The fractional order pid method with a fault tolerant module for road feeling control of steer-by-wire system. Mathematical Problems in Engineering, 2018, 1-12.

Yamaguchi, Y. and T. Murakami (2009). Adaptive control for virtual steering characteristics on electric vehicle using steer by wire system, IEEE Transactions on Industrial Electronics, Vol. 56, No. 5, pp. 1585-1594.

Yilei Zhang, Zibin Zheng Michael R. Lyu 2011, „BFTCloud: A Byzantine Fault Tolerance Framework for Voluntary-Resource Cloud Computing‟, IEEE 4th International Conference on Cloud Computing, pp. 444-451.

Zhou Ao, Shangguang Wang, Bo Cheng, Zibin Zheng, Fangchun Yang, Rong N. Chang, Michael R. Lyu Rajkumar Buyya 2017, „Cloud Service Reliability Enhancement via Virtual Machine Placement Optimization‟, IEEE Transactions on Service Computing, Iss. 06, Vol. 10, pp. 902-913

Ziqian Dong, Ning Liu Roberto Rojas-Cessa 2015, „Greedy scheduling of tasks with time constraints for energy-efficient cloud-computing data centers‟, Springer Journal of Cloud Computing: Advances, Systems and Applications, Vol. 4, No. 5.

Zong, C., Xiang, H., He, L., & Sha, F. (2012, April). Study on control method of dual-motor for steer-by-wire system. In 2012 2nd International Conference on Consumer Electronics, Communications and Networks (CECNet) (pp. 2890-2893). IEEE.

Author:

Senthil Kumar Durairajan is a distinguished product development manager and system engineering manager with over two decades of technical leadership experience in safety-critical product development across the automotive, avionics, medical, and locomotive industries. He holds certifications in INCOSE and NPDP and has delivered high-profile programs like ADAS and Autonomous Vehicles. Senthil specializes in system and software engineering, functional safety, cybersecurity, and AI. Currently a Senior Manager at Capgemini Engineering India, he continues to drive innovation in automotive engineering.