skip to main content

Title: Timing Debugging for Cyber-Physical Systems
This paper is concerned with the following question: Given a set of control tasks that are not schedulable, i.e., their required timing properties cannot be satisfied, what should be changed? While the real-time systems literature proposes many different schedulability analysis techniques, it surprisingly provides almost no guidelines on what should be changed to make a task set schedulable, when it is not. We show that when the tasks in question are control tasks, this timing debugging question in the context of cyber-physical systems (CPS) may be answered by exploiting the dynamics of the physical systems that these control tasks are expected to influence. Towards this, we study a very simple setup, viz., when a set of periodic tasks with implicit deadlines is not schedulable, by how much should the periods be changed in order to make the task set schedulable? Among the many ways in which the periods can be modified, our proposed strategy is to change the periods in a manner such that while the task set becomes schedulable, the poles of the closed-loop system experience the minimal shift. Since the poles influence the closed loop dynamics of the system, we thereby ensure that we obtain a system with the desired timing properties whose dynamics is very similar to the dynamics of the original (non-schedulable) system. We formulate this more » CPS timing debugging strategy as an optimization problem and illustrate it with a concrete example. « less
Authors:
; ; ; ;
Award ID(s):
1837337
Publication Date:
NSF-PAR ID:
10205882
Journal Name:
2021 Design, Automation Test in Europe Conference Exhibition (DATE)
Sponsoring Org:
National Science Foundation
More Like this
  1. Defense mechanisms against network-level attacks are commonly based on the use of cryptographic techniques, such as lengthy message authentication codes (MAC) that provide data integrity guarantees. However, such mechanisms require significant resources (both computational and network bandwidth), which prevents their continuous use in resource-constrained cyber-physical systems (CPS). Recently, it was shown how physical properties of controlled systems can be exploited to relax these stringent requirements for systems where sensor measurements and actuator commands are transmitted over a potentially compromised network; specifically, that merely intermittent use of data authentication (i.e., at occasional time points during system execution), can still provide strongmore »Quality-of-Control (QoC) guarantees even in the presence of false-data injection attacks, such as Man-in-the-Middle (MitM) attacks. Consequently, in this work, we focus on integrating security into existing resource-constrained CPS, in order to protect against MitM attacks on a system where a set of control tasks communicates over a real-time network with system sensors and actuators. We introduce a design-time methodology that incorporates requirements for QoC in the presence of attacks into end-to-end timing constraints for real-time control transactions, which include data acquisition and authentication, real-time network messages, and control tasks. This allows us to formulate a mixed integer linear programming-based method for direct synthesis of schedulable tasks and message parameters (i.e., deadlines and offsets) that do not violate timing requirements for the already deployed controllers, while adding a sufficient level of protection against network-based attacks; specifically, the synthesis method also provides suitable intermittent authentication policies that ensure the desired QoC levels under attack. To additionally reduce the security-related bandwidth overhead, we propose the use of cumulative message authentication at time instances when the integrity of messages from subsets of sensors should be ensured. Furthermore, we introduce a method for the opportunistic use of the remaining resources to further improve the overall QoC guarantees while ensuring system (i.e., task and message) schedulability. Finally, we demonstrate applicability and scalability of our methodology on synthetic automotive systems as well as a real-world automotive case-study.« less
  2. Obeid, I. ; Selesnik, I. ; Picone, J. (Ed.)
    The Neuronix high-performance computing cluster allows us to conduct extensive machine learning experiments on big data [1]. This heterogeneous cluster uses innovative scheduling technology, Slurm [2], that manages a network of CPUs and graphics processing units (GPUs). The GPU farm consists of a variety of processors ranging from low-end consumer grade devices such as the Nvidia GTX 970 to higher-end devices such as the GeForce RTX 2080. These GPUs are essential to our research since they allow extremely compute-intensive deep learning tasks to be executed on massive data resources such as the TUH EEG Corpus [2]. We use TensorFlow [3]more »as the core machine learning library for our deep learning systems, and routinely employ multiple GPUs to accelerate the training process. Reproducible results are essential to machine learning research. Reproducibility in this context means the ability to replicate an existing experiment – performance metrics such as error rates should be identical and floating-point calculations should match closely. Three examples of ways we typically expect an experiment to be replicable are: (1) The same job run on the same processor should produce the same results each time it is run. (2) A job run on a CPU and GPU should produce identical results. (3) A job should produce comparable results if the data is presented in a different order. System optimization requires an ability to directly compare error rates for algorithms evaluated under comparable operating conditions. However, it is a difficult task to exactly reproduce the results for large, complex deep learning systems that often require more than a trillion calculations per experiment [5]. This is a fairly well-known issue and one we will explore in this poster. Researchers must be able to replicate results on a specific data set to establish the integrity of an implementation. They can then use that implementation as a baseline for comparison purposes. A lack of reproducibility makes it very difficult to debug algorithms and validate changes to the system. Equally important, since many results in deep learning research are dependent on the order in which the system is exposed to the data, the specific processors used, and even the order in which those processors are accessed, it becomes a challenging problem to compare two algorithms since each system must be individually optimized for a specific data set or processor. This is extremely time-consuming for algorithm research in which a single run often taxes a computing environment to its limits. Well-known techniques such as cross-validation [5,6] can be used to mitigate these effects, but this is also computationally expensive. These issues are further compounded by the fact that most deep learning algorithms are susceptible to the way computational noise propagates through the system. GPUs are particularly notorious for this because, in a clustered environment, it becomes more difficult to control which processors are used at various points in time. Another equally frustrating issue is that upgrades to the deep learning package, such as the transition from TensorFlow v1.9 to v1.13, can also result in large fluctuations in error rates when re-running the same experiment. Since TensorFlow is constantly updating functions to support GPU use, maintaining an historical archive of experimental results that can be used to calibrate algorithm research is quite a challenge. This makes it very difficult to optimize the system or select the best configurations. The overall impact of all of these issues described above is significant as error rates can fluctuate by as much as 25% due to these types of computational issues. Cross-validation is one technique used to mitigate this, but that is expensive since you need to do multiple runs over the data, which further taxes a computing infrastructure already running at max capacity. GPUs are preferred when training a large network since these systems train at least two orders of magnitude faster than CPUs [7]. Large-scale experiments are simply not feasible without using GPUs. However, there is a tradeoff to gain this performance. Since all our GPUs use the NVIDIA CUDA® Deep Neural Network library (cuDNN) [8], a GPU-accelerated library of primitives for deep neural networks, it adds an element of randomness into the experiment. When a GPU is used to train a network in TensorFlow, it automatically searches for a cuDNN implementation. NVIDIA’s cuDNN implementation provides algorithms that increase the performance and help the model train quicker, but they are non-deterministic algorithms [9,10]. Since our networks have many complex layers, there is no easy way to avoid this randomness. Instead of comparing each epoch, we compare the average performance of the experiment because it gives us a hint of how our model is performing per experiment, and if the changes we make are efficient. In this poster, we will discuss a variety of issues related to reproducibility and introduce ways we mitigate these effects. For example, TensorFlow uses a random number generator (RNG) which is not seeded by default. TensorFlow determines the initialization point and how certain functions execute using the RNG. The solution for this is seeding all the necessary components before training the model. This forces TensorFlow to use the same initialization point and sets how certain layers work (e.g., dropout layers). However, seeding all the RNGs will not guarantee a controlled experiment. Other variables can affect the outcome of the experiment such as training using GPUs, allowing multi-threading on CPUs, using certain layers, etc. To mitigate our problems with reproducibility, we first make sure that the data is processed in the same order during training. Therefore, we save the data from the last experiment and to make sure the newer experiment follows the same order. If we allow the data to be shuffled, it can affect the performance due to how the model was exposed to the data. We also specify the float data type to be 32-bit since Python defaults to 64-bit. We try to avoid using 64-bit precision because the numbers produced by a GPU can vary significantly depending on the GPU architecture [11-13]. Controlling precision somewhat reduces differences due to computational noise even though technically it increases the amount of computational noise. We are currently developing more advanced techniques for preserving the efficiency of our training process while also maintaining the ability to reproduce models. In our poster presentation we will demonstrate these issues using some novel visualization tools, present several examples of the extent to which these issues influence research results on electroencephalography (EEG) and digital pathology experiments and introduce new ways to manage such computational issues.« less
  3. Existing design techniques for providing security guarantees against network-based attacks in cyber-physical systems (CPS) are based on continuous use of standard cryptographic tools to ensure data integrity. This creates an apparent conflict with common resource limitations in these systems, given that, for instance, lengthy message authentication codes (MAC) introduce significant overheads. We present a framework to ensure both timing guarantees for real-time network messages and Quality-of-Control (QoC) in the presence of network-based attacks. We exploit physical properties of controlled systems to relax constant integrity enforcement requirements, and show how the problem of feasibility testing of intermittently authenticated real-time messages canmore »be cast as a mixed integer linear programming problem. Besides scheduling a set of real-time messages with predefined authentication rates obtained from QoC requirements, we show how to optimally increase the overall system QoC while ensuring that all real-time messages are schedulable. Finally, we introduce an efficient runtime bandwidth allocation method, based on opportunistic scheduling, in order to improve QoC. We evaluate our framework on a standard benchmark designed for CAN bus, and show how an infeasible message set with strong security guarantees can be scheduled if dynamics of controlled systems are taken into account along with real-time requirements.« less
  4. We present a framework for deformable object manipulation that interleaves planning and control, enabling complex manipulation tasks without relying on high-fidelity modeling or simulation. The key question we address is when should we use planning and when should we use control to achieve the task? Planners are designed to find paths through complex configuration spaces, but for highly underactuated systems, such as deformable objects, achieving a specific configuration is very difficult even with high-fidelity models. Conversely, controllers can be designed to achieve specific configurations, but they can be trapped in undesirable local minima owing to obstacles. Our approach consists ofmore »three components: (1) a global motion planner to generate gross motion of the deformable object; (2) a local controller for refinement of the configuration of the deformable object; and (3) a novel deadlock prediction algorithm to determine when to use planning versus control. By separating planning from control we are able to use different representations of the deformable object, reducing overall complexity and enabling efficient computation of motion. We provide a detailed proof of probabilistic completeness for our planner, which is valid despite the fact that our system is underactuated and we do not have a steering function. We then demonstrate that our framework is able to successfully perform several manipulation tasks with rope and cloth in simulation, which cannot be performed using either our controller or planner alone. These experiments suggest that our planner can generate paths efficiently, taking under a second on average to find a feasible path in three out of four scenarios. We also show that our framework is effective on a 16-degree-of-freedom physical robot, where reachability and dual-arm constraints make the planning more difficult.« less
  5. Physical interaction with tools is ubiquitous in functional activities of daily living. While tool use is considered a hallmark of human behavior, how humans control such physical interactions is still poorly understood. When humans perform a motor task, it is commonly suggested that the central nervous system coordinates the musculo-skeletal system to minimize muscle effort. In this paper, we tested if this notion holds true for motor tasks that involve physical interaction. Specifically, we investigated whether humans minimize muscle forces to control physical interaction with a circular kinematic constraint. Using a simplified arm model, we derived three predictions for howmore »humans should behave if they were minimizing muscular effort to perform the task. First, we predicted that subjects would exert workless, radial forces on the constraint. Second, we predicted that the muscles would be deactivated when they could not contribute to work. Third, we predicted that when moving very slowly along the constraint, the pattern of muscle activity would not differ between clockwise (CW) and counterclockwise (CCW) motions. To test these predictions, we instructed human subjects to move a robot handle around a virtual, circular constraint at a constant tangential velocity. To reduce the effect of forces that might arise from incomplete compensation of neuro-musculo-skeletal dynamics, the target tangential speed was set to an extremely slow pace (~1 revolution every 13.3 seconds). Ultimately, the results of human experiment did not support the predictions derived from our model of minimizing muscular effort. While subjects did exert workless forces, they did not deactivate muscles as predicted. Furthermore, muscle activation patterns differed between CW and CCW motions about the constraint. These findings demonstrate that minimizing muscle effort is not a significant factor in human performance of this constrained-motion task. Instead, the central nervous system likely prioritizes reducing other costs, such as computational effort, over muscle effort to control physical interactions.« less