Motivated by the increasingly wide adoption of realtime workload with self-suspending behaviors, and the relevance of mechanisms to handle mutually-exclusive shared resources, this paper takes a new look at locking protocols for self-suspending tasks under uniprocessor fixed-priority scheduling. Pitfalls when integrating the widely-adopted Stack Resource Policy (SRP) with self-suspending tasks are firstly illustrated, and then a new finegrained SRP analysis is presented. Next, a new locking protocol, named SRP-SS, is proposed to overcome the limitations of the original SRP. The SRP-SS is a generalization of the SRP to cope with the specificities of self-suspending tasks. It therefore reduces to the SRP under some configurations and hence theoretically dominates the SRP. It also ensures backward compatibility for applications developed specifically for the SRP. The SRP-SS comes with its own schedulability analysis and configuration algorithm. The performances of the SRP and SRP-SS are finally studied by means of large-scale schedulability experiments.
{"title":"The SRP Resource Sharing Protocol for Self-Suspending Tasks","authors":"Geoffrey Nelissen, Alessandro Biondi","doi":"10.1109/RTSS.2018.00051","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00051","url":null,"abstract":"Motivated by the increasingly wide adoption of realtime workload with self-suspending behaviors, and the relevance of mechanisms to handle mutually-exclusive shared resources, this paper takes a new look at locking protocols for self-suspending tasks under uniprocessor fixed-priority scheduling. Pitfalls when integrating the widely-adopted Stack Resource Policy (SRP) with self-suspending tasks are firstly illustrated, and then a new finegrained SRP analysis is presented. Next, a new locking protocol, named SRP-SS, is proposed to overcome the limitations of the original SRP. The SRP-SS is a generalization of the SRP to cope with the specificities of self-suspending tasks. It therefore reduces to the SRP under some configurations and hence theoretically dominates the SRP. It also ensures backward compatibility for applications developed specifically for the SRP. The SRP-SS comes with its own schedulability analysis and configuration algorithm. The performances of the SRP and SRP-SS are finally studied by means of large-scale schedulability experiments.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125033309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For real-time computing systems, energy efficiency and reliability are two primary design concerns. In this research work, we study the problem of enhanced energy-aware standbysparing for fixed-priority (FP) hard real-time systems under reliability requirement. The standby-sparing system adopts a primary processor and a spare processor to provide fault tolerance for both permanent and transient faults. In order to keep the energy consumption for such kind of systems under control, we explore enhanced fixed-priority scheduling schemes to minimize the overlapped concurrent executions of the workloads on the primary processor and on the spare processor, enabling energy savings. Moreover, efficient online scheduling techniques are under development to boost the energy savings during runtime while preserving the system reliability.
{"title":"Work-in-Progress: Enhanced Energy-Aware Standby-Sparing Techniques for Fixed-Priority Hard Real-Time Systems","authors":"Linwei Niu, Jonathan Musselwhite, Wei Li","doi":"10.1109/RTSS.2018.00031","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00031","url":null,"abstract":"For real-time computing systems, energy efficiency and reliability are two primary design concerns. In this research work, we study the problem of enhanced energy-aware standbysparing for fixed-priority (FP) hard real-time systems under reliability requirement. The standby-sparing system adopts a primary processor and a spare processor to provide fault tolerance for both permanent and transient faults. In order to keep the energy consumption for such kind of systems under control, we explore enhanced fixed-priority scheduling schemes to minimize the overlapped concurrent executions of the workloads on the primary processor and on the spare processor, enabling energy savings. Moreover, efficient online scheduling techniques are under development to boost the energy savings during runtime while preserving the system reliability.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125814732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Aakash Soni, Xiaoting Li, Jean-Luc Scharbarg, C. Fraboul
Avionics Full Duplex switched Ethernet (AFDX) is the de facto standard for the transmission of critical avionics flows. It is a specific switched Ethernet solution based on First-in First-out (FIFO) scheduling. Worst-case traversal time (WCTT) analysis is mandatory for such flows, since timing constraints have to be guaranteed. A classical approach in this context is Network Calculus (NC). However, NC introduces some pessimism in the WCTT computation. Moreover, the worst-case often corresponds to very rare scenarios. Thus, the network architecture is most of the time lightly loaded. Typically, less than 10 % of the available bandwidth is used for the transmission of avionics lows on an AFDX network embedded in an aircraft. One solution to improve the utilization of the network is to introduce Quality of Service (QoS) mechanisms. Deficit Round Robin (DRR) is such a mechanism and it is envisioned for future avionics networks. A WCTT analysis has been proposed for DRR. It is based on NC. It doesn't make any assumption on the scheduling of flows by end systems. The first contribution of this paper is to identify sources of pessimism of this approach and to propose an improved solution which removes part of this pessimism. The second contribution is to show how the scheduling of flows can be integrated in this optimized DRR approach, thanks to offsets. An evaluation on a realistic case study shows that both contributions bring significantly tighter bounds on worst-case latencies.
{"title":"Optimizing Network Calculus for Switched Ethernet Network with Deficit Round Robin","authors":"Aakash Soni, Xiaoting Li, Jean-Luc Scharbarg, C. Fraboul","doi":"10.1109/RTSS.2018.00046","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00046","url":null,"abstract":"Avionics Full Duplex switched Ethernet (AFDX) is the de facto standard for the transmission of critical avionics flows. It is a specific switched Ethernet solution based on First-in First-out (FIFO) scheduling. Worst-case traversal time (WCTT) analysis is mandatory for such flows, since timing constraints have to be guaranteed. A classical approach in this context is Network Calculus (NC). However, NC introduces some pessimism in the WCTT computation. Moreover, the worst-case often corresponds to very rare scenarios. Thus, the network architecture is most of the time lightly loaded. Typically, less than 10 % of the available bandwidth is used for the transmission of avionics lows on an AFDX network embedded in an aircraft. One solution to improve the utilization of the network is to introduce Quality of Service (QoS) mechanisms. Deficit Round Robin (DRR) is such a mechanism and it is envisioned for future avionics networks. A WCTT analysis has been proposed for DRR. It is based on NC. It doesn't make any assumption on the scheduling of flows by end systems. The first contribution of this paper is to identify sources of pessimism of this approach and to propose an improved solution which removes part of this pessimism. The second contribution is to show how the scheduling of flows can be integrated in this optimized DRR approach, thanks to offsets. An evaluation on a realistic case study shows that both contributions bring significantly tighter bounds on worst-case latencies.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125443159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Habeeb Olufowobi, Gedare Bloom, C. Young, Joseph Zambreno
Security of vehicular networks has often been an afterthought since they are designed traditionally to be a closed system. An attack could lead to catastrophic effect which may include loss of human life or severe injury to the driver and passengers of the vehicle. In this paper, we propose a novel algorithm to extract the real-time model of the controller area network (CAN) and develop a specification-based intrusion detection system (IDS) using anomaly-based supervised learning with the real-time model as input. We evaluate IDS performance with real CAN logs collected from a sedan car.
{"title":"Work-in-Progress: Real-Time Modeling for Intrusion Detection in Automotive Controller Area Network","authors":"Habeeb Olufowobi, Gedare Bloom, C. Young, Joseph Zambreno","doi":"10.1109/RTSS.2018.00030","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00030","url":null,"abstract":"Security of vehicular networks has often been an afterthought since they are designed traditionally to be a closed system. An attack could lead to catastrophic effect which may include loss of human life or severe injury to the driver and passengers of the vehicle. In this paper, we propose a novel algorithm to extract the real-time model of the controller area network (CAN) and develop a specification-based intrusion detection system (IDS) using anomaly-based supervised learning with the real-time model as input. We evaluate IDS performance with real CAN logs collected from a sedan car.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"148 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122150231","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nicola Capodieci, R. Cavicchioli, M. Bertogna, Aingara Paramakuru
Modern automotive-grade embedded computing platforms feature high-performance Graphics Processing Units (GPUs) to support the massively parallel processing power needed for next-generation autonomous driving applications (e.g., Deep Neural Network (DNN) inference, sensor fusion, path planning, etc). As these workload-intensive activities are pushed to higher criticality levels, there is a stronger need for more predictable scheduling algorithms that are able to guarantee predictability without overly sacrificing GPU utilization. Unfortunately, the real-rime literature on GPU scheduling mostly considered limited (or null) preemption capabilities, while previous efforts in broader domains were often based on programming models and APIs that were not designed to support the real-rime requirements of recurring workloads. In this paper, we present the design of a prototype real-time scheduler for GPU activities on an embedded System on a Chip (SoC) featuring a cutting edge GPU architecture by NVIDIA adopted in the autonomous driving domain. The scheduler runs as a software partition on top of the NVIDIA hypervisor, and it leverages latest generation architectural features, such as pixel-level preemption and threadlevel preemption. Such a design allowed us to implement and test a preemptive Earliest Deadline First (EDF) scheduler for GPU tasks providing bandwidth isolations by means of a Constant Bandwidth Server (CBS). Our work involved investigating alternative programming models for compute APIs, allowing us to characterize CPU-to-GPU command submission with more detailed scheduling information. A detailed experimental characterization is presented to show the significant schedulability improvement of recurring real-time GPU tasks.
{"title":"Deadline-Based Scheduling for GPU with Preemption Support","authors":"Nicola Capodieci, R. Cavicchioli, M. Bertogna, Aingara Paramakuru","doi":"10.1109/RTSS.2018.00021","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00021","url":null,"abstract":"Modern automotive-grade embedded computing platforms feature high-performance Graphics Processing Units (GPUs) to support the massively parallel processing power needed for next-generation autonomous driving applications (e.g., Deep Neural Network (DNN) inference, sensor fusion, path planning, etc). As these workload-intensive activities are pushed to higher criticality levels, there is a stronger need for more predictable scheduling algorithms that are able to guarantee predictability without overly sacrificing GPU utilization. Unfortunately, the real-rime literature on GPU scheduling mostly considered limited (or null) preemption capabilities, while previous efforts in broader domains were often based on programming models and APIs that were not designed to support the real-rime requirements of recurring workloads. In this paper, we present the design of a prototype real-time scheduler for GPU activities on an embedded System on a Chip (SoC) featuring a cutting edge GPU architecture by NVIDIA adopted in the autonomous driving domain. The scheduler runs as a software partition on top of the NVIDIA hypervisor, and it leverages latest generation architectural features, such as pixel-level preemption and threadlevel preemption. Such a design allowed us to implement and test a preemptive Earliest Deadline First (EDF) scheduler for GPU tasks providing bandwidth isolations by means of a Constant Bandwidth Server (CBS). Our work involved investigating alternative programming models for compute APIs, allowing us to characterize CPU-to-GPU command submission with more detailed scheduling information. A detailed experimental characterization is presented to show the significant schedulability improvement of recurring real-time GPU tasks.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122205240","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Processing and analyzing big data sets updated in real time in an increasing number of applications such as severe weather prediction and particle-physics experiments require the computational power of extreme-scale high-performance computing (HPC) systems. To address the scheduling of massive task/thread sets on these extreme-scale systems, current strategies rely on improving centralized, distributed, and parallel scheduling algorithms as well as virtualization developed for HPC systems which aim to reduce the makespan and balance the load among the computing nodes in these systems. However, these HPC schedulers provide no guarantees on meeting timing constraints such as deadlines that are required in an increasing number of these real-time science workflows. This paper describes a new project which departs from this established trend of best-effort scheduling of large-scale HPC Message Passing Interface (MPI) tasks and ensemble workloads found in fine-grain many-task computing (MTC) applications. The new approach brings real-time scheduling to address the demands of real-time science workloads. This new framework abstracts information about the tasks or threads, and continuously dispatch this workload to meet deadlines and other timing constraints associated with individual tasks or groups of tasks in extreme-scale HPC systems to reduce execution time and energy consumption. This paper introduces deadline-based scheduling in the tasking programming model.
{"title":"Work-in-Progress: Incorporating Deadline-Based Scheduling in Tasking Programming Model for Extreme-Scale Parallel Computing","authors":"A. Cheng, Panruo Wu","doi":"10.1109/RTSS.2018.00022","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00022","url":null,"abstract":"Processing and analyzing big data sets updated in real time in an increasing number of applications such as severe weather prediction and particle-physics experiments require the computational power of extreme-scale high-performance computing (HPC) systems. To address the scheduling of massive task/thread sets on these extreme-scale systems, current strategies rely on improving centralized, distributed, and parallel scheduling algorithms as well as virtualization developed for HPC systems which aim to reduce the makespan and balance the load among the computing nodes in these systems. However, these HPC schedulers provide no guarantees on meeting timing constraints such as deadlines that are required in an increasing number of these real-time science workflows. This paper describes a new project which departs from this established trend of best-effort scheduling of large-scale HPC Message Passing Interface (MPI) tasks and ensemble workloads found in fine-grain many-task computing (MTC) applications. The new approach brings real-time scheduling to address the demands of real-time science workloads. This new framework abstracts information about the tasks or threads, and continuously dispatch this workload to meet deadlines and other timing constraints associated with individual tasks or groups of tasks in extreme-scale HPC systems to reduce execution time and energy consumption. This paper introduces deadline-based scheduling in the tasking programming model.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127910442","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Sandeep Kumar Bijinemula, Aaron Willcock, Thidapat Chantem, N. Fisher
Engine-triggered tasks are real-time tasks that are released when the crankshaft in an engine completes a rotation, which depends on the angular speed and acceleration of the crankshaft itself. In addition, the execution time of an engine-triggered task depends on the speed of the crankshaft. Tasks whose execution times depend on a variable period are referred to as adaptive-variable rate (AVR) tasks. Existing techniques to calculate the worst-case demand of AVR tasks are either inexact or computationally intractable. In this paper, we transform the problem of finding the worst-case demand of AVR tasks over a given time interval into a variant of the knapsack problem to efficiently find the exact solution. We then propose a framework to systematically reduce the search space associated with finding the worst-case demand of AVR tasks. Experimental results reveal that our approach is at least 10 times faster, with an average runtime improvement of 146 times, for randomly generated tasksets when compared to the state-of-the-art technique.
{"title":"An Efficient Knapsack-Based Approach for Calculating the Worst-Case Demand of AVR Tasks","authors":"Sandeep Kumar Bijinemula, Aaron Willcock, Thidapat Chantem, N. Fisher","doi":"10.1109/RTSS.2018.00053","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00053","url":null,"abstract":"Engine-triggered tasks are real-time tasks that are released when the crankshaft in an engine completes a rotation, which depends on the angular speed and acceleration of the crankshaft itself. In addition, the execution time of an engine-triggered task depends on the speed of the crankshaft. Tasks whose execution times depend on a variable period are referred to as adaptive-variable rate (AVR) tasks. Existing techniques to calculate the worst-case demand of AVR tasks are either inexact or computationally intractable. In this paper, we transform the problem of finding the worst-case demand of AVR tasks over a given time interval into a variant of the knapsack problem to efficiently find the exact solution. We then propose a framework to systematically reduce the search space associated with finding the worst-case demand of AVR tasks. Experimental results reveal that our approach is at least 10 times faster, with an average runtime improvement of 146 times, for randomly generated tasksets when compared to the state-of-the-art technique.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115591064","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
For a set of real-time tasks that have mixed preference of being executed at early or late times before their deadlines, we have recently studied both earliest-deadline based and fixed-priority preference-oriented (PO) scheduling algorithms for uniprocessor systems. In this work, focusing on multiprocessor real-time systems, we study the foundational guidelines to design partition-based PO scheduling algorithms for tasks with mixed preference requirements. In particular, through a concrete example, we illustrate that the harmonicity of tasks' periods should be incorporated when making scheduling decisions in addition to their execution preferences to obtain favorable schedules that better fulfill tasks' preference requirements. Based on such guidelines, we design a period-aware preference-oriented (PAPO) partitioned scheduling algorithm and discuss several variations by considering harmonicity as well as utilization of tasks.
{"title":"Work-in-Progress: Preference-Oriented Scheduling in Multiprocessor Real-Time Systems","authors":"Qin Xia, Dakai Zhu, Hakan Aydin","doi":"10.1109/RTSS.2018.00023","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00023","url":null,"abstract":"For a set of real-time tasks that have mixed preference of being executed at early or late times before their deadlines, we have recently studied both earliest-deadline based and fixed-priority preference-oriented (PO) scheduling algorithms for uniprocessor systems. In this work, focusing on multiprocessor real-time systems, we study the foundational guidelines to design partition-based PO scheduling algorithms for tasks with mixed preference requirements. In particular, through a concrete example, we illustrate that the harmonicity of tasks' periods should be incorporated when making scheduling decisions in addition to their execution preferences to obtain favorable schedules that better fulfill tasks' preference requirements. Based on such guidelines, we design a period-aware preference-oriented (PAPO) partitioned scheduling algorithm and discuss several variations by considering harmonicity as well as utilization of tasks.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"2012 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125891431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Guang Wang, Xiaoyan Xie, Fan Zhang, Yunhuai Liu, Desheng Zhang
We are witnessing a rapid growth of electrified vehicles because of the ever-increasing concerns over urban air quality and energy security. Compared with other electric vehicles, electric buses have not yet been prevailingly adopted worldwide due to the high owning and operating costs, long charging time, and the uneven distribution of charging facilities. Moreover, the highly dynamic environment factors such as the unpredictable traffic congestions, different passenger demands, and even changing weather, can significantly affect electric bus charging efficiency and potentially hinder further development of large-scale electric bus fleets. To deal with these issues, in this paper, we first analyze a real-world dataset including massive data from 16,359 electric buses, 1,400 bus lines and 5,562 bus stops, which is obtained from the Chinese city Shenzhen, who has the first and the largest full electric bus network for public transit. Then we investigate the electric bus network to understand its operating and charging patterns, and further verify the feasibility and necessity of a real-time charging scheduling. With such understanding, we design bCharge, a real-time charging scheduling system based on Markov Decision Process to reduce the overall charging and operating costs for city-scale electric bus fleets, taking the time-variant electricity pricing into account. To show the effectiveness of bCharge, we implement it with the real-world streaming dataset from Shenzhen, which includes GPS data of the electric bus fleet, the bus lines and stops data, coupled with the 376 electric bus charging stations data. The evaluation results show that bCharge can dramatically reduce the charging cost by 23.7% and 12.8% electricity usage simultaneously.
{"title":"bCharge: Data-Driven Real-Time Charging Scheduling for Large-Scale Electric Bus Fleets","authors":"Guang Wang, Xiaoyan Xie, Fan Zhang, Yunhuai Liu, Desheng Zhang","doi":"10.1109/RTSS.2018.00015","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00015","url":null,"abstract":"We are witnessing a rapid growth of electrified vehicles because of the ever-increasing concerns over urban air quality and energy security. Compared with other electric vehicles, electric buses have not yet been prevailingly adopted worldwide due to the high owning and operating costs, long charging time, and the uneven distribution of charging facilities. Moreover, the highly dynamic environment factors such as the unpredictable traffic congestions, different passenger demands, and even changing weather, can significantly affect electric bus charging efficiency and potentially hinder further development of large-scale electric bus fleets. To deal with these issues, in this paper, we first analyze a real-world dataset including massive data from 16,359 electric buses, 1,400 bus lines and 5,562 bus stops, which is obtained from the Chinese city Shenzhen, who has the first and the largest full electric bus network for public transit. Then we investigate the electric bus network to understand its operating and charging patterns, and further verify the feasibility and necessity of a real-time charging scheduling. With such understanding, we design bCharge, a real-time charging scheduling system based on Markov Decision Process to reduce the overall charging and operating costs for city-scale electric bus fleets, taking the time-variant electricity pricing into account. To show the effectiveness of bCharge, we implement it with the real-world streaming dataset from Shenzhen, which includes GPS data of the electric bus fleet, the bus lines and stops data, coupled with the 376 electric bus charging stations data. The evaluation results show that bCharge can dramatically reduce the charging cost by 23.7% and 12.8% electricity usage simultaneously.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128424874","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Soroush Bateni, Husheng Zhou, Yuankun Zhu, Cong Liu
The revolution of deep neural networks (DNNs) is enabling dramatically better autonomy in autonomous driving. However, it is not straightforward to simultaneously achieve both timing predictability (i.e., meeting job latency requirements) and energy efficiency that are essential for any DNN-based autonomous driving system, as they represent two (often) conflicting goals. In this paper, we propose PredJoule, a timing-predictable energy optimization framework for running DNN workloads in a GPU-enabled automotive system. PredJoule achieves both latency guarantees and energy efficiency through a layer-aware design that explores specific performance and energy characteristics of different layers within the same neural network. We implement and evaluate PredJoule on the automotive-specific NVIDIA Jetson TX2 platform for five state-of-the-art DNN models with both high and low variance latency requirements. Experiments show that PredJoule rarely violates job deadlines, and can improve energy by 65% on average compared to five existing approaches and 68% compared to an energy-oriented approach.
{"title":"PredJoule: A Timing-Predictable Energy Optimization Framework for Deep Neural Networks","authors":"Soroush Bateni, Husheng Zhou, Yuankun Zhu, Cong Liu","doi":"10.1109/RTSS.2018.00020","DOIUrl":"https://doi.org/10.1109/RTSS.2018.00020","url":null,"abstract":"The revolution of deep neural networks (DNNs) is enabling dramatically better autonomy in autonomous driving. However, it is not straightforward to simultaneously achieve both timing predictability (i.e., meeting job latency requirements) and energy efficiency that are essential for any DNN-based autonomous driving system, as they represent two (often) conflicting goals. In this paper, we propose PredJoule, a timing-predictable energy optimization framework for running DNN workloads in a GPU-enabled automotive system. PredJoule achieves both latency guarantees and energy efficiency through a layer-aware design that explores specific performance and energy characteristics of different layers within the same neural network. We implement and evaluate PredJoule on the automotive-specific NVIDIA Jetson TX2 platform for five state-of-the-art DNN models with both high and low variance latency requirements. Experiments show that PredJoule rarely violates job deadlines, and can improve energy by 65% on average compared to five existing approaches and 68% compared to an energy-oriented approach.","PeriodicalId":294784,"journal":{"name":"2018 IEEE Real-Time Systems Symposium (RTSS)","volume":"PP 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126421143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：