Odessa Enabling Interactive Perception Applications on Mobile Devices
•
1 like•75 views
This document summarizes the Odessa system for enabling interactive perception applications on mobile devices through offloading and parallelism. Odessa uses a greedy approach to dynamically determine how to best distribute computation between a mobile device and more powerful servers. It considers factors like input data variability, mobile platform capabilities, and network performance. The system introduces low overhead through a lightweight profiler and decision engine. Evaluation shows it can minimize response time while achieving high throughput and accuracy for computer vision applications.
Report
Share
Report
Share
Download to read offline
Recommended
Hidden Markov Models for Abnormal Event Processing in Transportation Data Str...
The document discusses using Hidden Markov Models (HMM) to process transportation data streams and identify abnormal events in real-time. An Event-Driven Architecture is used to decouple the computationally expensive HMM calculations from real-time event processing. The HMM is used to determine time ranges for normal events, and any incoming events outside this range are flagged as abnormal. Experimental results show the approach can accurately detect abnormal events in transportation data streams. Future work involves improving the HMM to detect more specific event types like accidents, and using Map-Reduce to optimize the HMM computations.
Presentation UCAMI Congress 2016
A GIS water management system was developed using free and open source software as a strategic element for smart cities. The existing proprietary solution had limitations including a generic data model, restrictions on external system connections, and limitations on publishing information. The migration process involved redesigning the data model, creating geometric and thematic validation systems, matching models and importing data, connecting to subsystems, and developing client functionality. The resulting open source solution provides flexibility, scalability, and interoperability to address smart city needs.
Timeline: An Operating System Abstraction for Time-Aware Applications
This document summarizes a research paper about Timeline, an operating system abstraction that allows time-aware applications to be aware of timing uncertainty in distributed systems. The abstraction introduces the concept of Quality-of-Time (QoT) to convey uncertainty. It presents a QoT architecture that uses timelines maintained as trees and bindings as linked lists to coordinate time synchronization across nodes based on application requirements. The implementation uses a data distribution service and synchronization services. Evaluation results show the approach enables choreographed scheduling while balancing performance and resource usage.
FlumeJava: Easy, Efficient Data-Parallel Pipelines
FlumeJava is a Java library that provides high-level abstractions for writing efficient data-parallel pipelines that are executed using MapReduce. It allows defining parallel collections and operations on those collections in a deferred manner, building a directed acyclic graph. An optimizer then transforms the graph by fusing operations together into single MapReduce jobs to improve efficiency. FlumeJava is used widely at Google to simplify developing, testing, and running data-parallel applications that are ultimately executed on MapReduce.
Resume
Hongwei Xie is pursuing a Master's degree in Information Systems Management at Carnegie Mellon University with a 3.7 GPA. He received a B.S. in Information Systems Management from Beijing Institute of Technology with a 3.82 GPA. He has experience in programming languages like Java, SQL, Python and Bash. He has worked as a Software Engineer Intern at CleanRobotics where he designed databases and automated processes for their trash robot. His projects include Twitter data analytics on AWS and a local crime reporting application. He has received honors for his work in mathematical modeling competitions.
SFScon21 - Gianluca Antonacci - Optimal motorway speed management. a real-tim...
SFScon21 - Gianluca Antonacci - Optimal motorway speed management. a real-tim...South Tyrol Free Software Conference
A real-time algorithm for calculating the best flow condition
We present a method for calculating the optimal motorway speed based on real-time measured data. On the basis of the vehicle flow and the transit speed measured at high frequency (1 minute), an algorithm was developed based on the concept of a “state machine”, which in real time calculates the optimal transit speed. This information can be immediately communicated to drivers via variable message signs. The optimal speed is the one that avoids or reduces traffic jams and ensures maximum flow of a road section.
The system is currently being tested by the A22 Brenner motorway as part of the European LIFE+ project BrennerLEC, but it is conceptually applicable to any motorway equipped with a measurement infrastructure for vehicular transit and signalling to users.Hierarchical decentralized network reconfiguration for smart distribution sys...
Hierarchical decentralized network reconfiguration for smart distribution sys...Pvrtechnologies Nellore
The document presents a hierarchical decentralized network reconfiguration approach to minimize power losses in smart distribution systems. It applies the proposed approach to four test distribution systems and compares the results to existing centralized and harmony search methods. The decentralized approach can achieve similar results as other methods with significantly reduced computation time. It also studies the impacts of time-varying loads and faults through dynamic network reconfiguration on a 118-bus test system. Simulation results show the optimal configuration for each operating period can significantly reduce energy losses.Covid 19 monitor
This 3 sentence summary provides the key details about a Covid-19 monitoring tool:
The tool crawls websites to obtain the latest Covid-19 data, which it then uses to generate visual analysis like graphs and charts to help data scientists, statisticians, governments, and others visualize the situation and future predictions. It is built using technologies like Python, Flask, Beautiful Soup and Pandas and hosted on Heroku, and has potential for extensions like notifications and predicting how environmental factors may impact the spread of Covid-19.
Recommended
Hidden Markov Models for Abnormal Event Processing in Transportation Data Str...
The document discusses using Hidden Markov Models (HMM) to process transportation data streams and identify abnormal events in real-time. An Event-Driven Architecture is used to decouple the computationally expensive HMM calculations from real-time event processing. The HMM is used to determine time ranges for normal events, and any incoming events outside this range are flagged as abnormal. Experimental results show the approach can accurately detect abnormal events in transportation data streams. Future work involves improving the HMM to detect more specific event types like accidents, and using Map-Reduce to optimize the HMM computations.
Presentation UCAMI Congress 2016
A GIS water management system was developed using free and open source software as a strategic element for smart cities. The existing proprietary solution had limitations including a generic data model, restrictions on external system connections, and limitations on publishing information. The migration process involved redesigning the data model, creating geometric and thematic validation systems, matching models and importing data, connecting to subsystems, and developing client functionality. The resulting open source solution provides flexibility, scalability, and interoperability to address smart city needs.
Timeline: An Operating System Abstraction for Time-Aware Applications
This document summarizes a research paper about Timeline, an operating system abstraction that allows time-aware applications to be aware of timing uncertainty in distributed systems. The abstraction introduces the concept of Quality-of-Time (QoT) to convey uncertainty. It presents a QoT architecture that uses timelines maintained as trees and bindings as linked lists to coordinate time synchronization across nodes based on application requirements. The implementation uses a data distribution service and synchronization services. Evaluation results show the approach enables choreographed scheduling while balancing performance and resource usage.
FlumeJava: Easy, Efficient Data-Parallel Pipelines
FlumeJava is a Java library that provides high-level abstractions for writing efficient data-parallel pipelines that are executed using MapReduce. It allows defining parallel collections and operations on those collections in a deferred manner, building a directed acyclic graph. An optimizer then transforms the graph by fusing operations together into single MapReduce jobs to improve efficiency. FlumeJava is used widely at Google to simplify developing, testing, and running data-parallel applications that are ultimately executed on MapReduce.
Resume
Hongwei Xie is pursuing a Master's degree in Information Systems Management at Carnegie Mellon University with a 3.7 GPA. He received a B.S. in Information Systems Management from Beijing Institute of Technology with a 3.82 GPA. He has experience in programming languages like Java, SQL, Python and Bash. He has worked as a Software Engineer Intern at CleanRobotics where he designed databases and automated processes for their trash robot. His projects include Twitter data analytics on AWS and a local crime reporting application. He has received honors for his work in mathematical modeling competitions.
SFScon21 - Gianluca Antonacci - Optimal motorway speed management. a real-tim...
SFScon21 - Gianluca Antonacci - Optimal motorway speed management. a real-tim...South Tyrol Free Software Conference
A real-time algorithm for calculating the best flow condition
We present a method for calculating the optimal motorway speed based on real-time measured data. On the basis of the vehicle flow and the transit speed measured at high frequency (1 minute), an algorithm was developed based on the concept of a “state machine”, which in real time calculates the optimal transit speed. This information can be immediately communicated to drivers via variable message signs. The optimal speed is the one that avoids or reduces traffic jams and ensures maximum flow of a road section.
The system is currently being tested by the A22 Brenner motorway as part of the European LIFE+ project BrennerLEC, but it is conceptually applicable to any motorway equipped with a measurement infrastructure for vehicular transit and signalling to users.Hierarchical decentralized network reconfiguration for smart distribution sys...
Hierarchical decentralized network reconfiguration for smart distribution sys...Pvrtechnologies Nellore
The document presents a hierarchical decentralized network reconfiguration approach to minimize power losses in smart distribution systems. It applies the proposed approach to four test distribution systems and compares the results to existing centralized and harmony search methods. The decentralized approach can achieve similar results as other methods with significantly reduced computation time. It also studies the impacts of time-varying loads and faults through dynamic network reconfiguration on a 118-bus test system. Simulation results show the optimal configuration for each operating period can significantly reduce energy losses.Covid 19 monitor
This 3 sentence summary provides the key details about a Covid-19 monitoring tool:
The tool crawls websites to obtain the latest Covid-19 data, which it then uses to generate visual analysis like graphs and charts to help data scientists, statisticians, governments, and others visualize the situation and future predictions. It is built using technologies like Python, Flask, Beautiful Soup and Pandas and hosted on Heroku, and has potential for extensions like notifications and predicting how environmental factors may impact the spread of Covid-19.
Cloud Migration Strategy and Best Practices
Analyze key aspects to be considered before embarking on your cloud journey. The presentation outlines the strategies, approach, and choices that need to be made, to ensure a smooth transition to the cloud.
cloud computing.pptx fundamentals and deployment models
Cloud computing allows sharing of data, storage, and access to computer resources over the internet. It has three main service models: Infrastructure as a Service (IaaS) provides basic services and hardware that users maintain; Platform as a Service (PaaS) provides software, hardware, and platforms automatically; and Software as a Service (SaaS) provides transparent access to software and hardware. Cloud computing also has four deployment models: public, community, private, and hybrid clouds. While cloud computing provides benefits, it also faces challenges regarding privacy, compliance with regulations, security, and potential abuse.
Presented by Ahmed Abdulhakim Al-Absi - Scaling map reduce applications acro...
Scaling map reduce applications across hybrid clouds to meet soft deadlines - By Michael Mattess, Rodrigo N. Calheiros, and Rajkumar Buyya, Proceedings of the 27th IEEE International Conference on Advanced Information Networking and Applications (AINA 2013, IEEE CS Press, USA), Barcelona, Spain, March 25-28, 2013.
RECAP Project Overview
This a RECAP project overview slide deck prepared by Thang Le Duc (UMU), P-O Östberg (UMU) and Tomas Brännström (Tieto). It starts with an introduction and continues with a section on challenges for a self-orchestrated, self-remediated cloud system. It then presents the RECAP vision and use cases and finishes with a conclusion.
Trafficshifting: Avoiding Disasters & Improving Performance at Scale
LinkedIn uses traffic shifting techniques to avoid disasters and improve performance. They shift traffic between their datacenter fabrics and point of presence (PoP) locations using sticky routing and DNS steering. This allows them to perform disaster recovery testing and mitigate outages. Specifically for the Asia Pacific region, they overcame challenges such as fiber cuts by failing traffic out of affected PoP locations. Their techniques improved performance and user experience for both IPv4 and IPv6 users.
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice by José I. Benedetto Andrés Neyem Jaime Navón Guillermo Valenzuela. MobileSoft 2017, Buenos Aires.
APRICOT 2017: Trafficshifting: Avoiding Disasters & Improving Performance at ...
LinkedIn serves traffic for its 467 million members from four data centers and multiple PoPs spread geographically around the world. Serving live traffic from from many places at the same time has taken us from a disaster recovery model to a disaster avoidance model where we can take an unhealthy data center or PoP out of rotation and redistribute its traffic to a healthy one within minutes, with virtually no visible impact to users. The geographical distribution of our infrastructure also allows us to optimize the end-user's experience by geo routing users to the best possible PoP and datacenter.
This talk provide details on how LinkedIn shifts traffic between its PoPs and data centers to provide the best possible performance and availability for its members. We will also touch on the complexities of performance in APAC, how IPv6 is helping our members and how LinkedIn stress tests data centers verify its disaster recovery capabilities.
5G-USA-Telemetry
1) Google has built one of the fastest and most capable network infrastructures over the past 15+ years through innovations like global caching, software defined networking, and virtualizing the physical network.
2) Telemetry and analytics are needed in large data center networks to perform network modeling, configuration verification, and fault isolation given their complexity with thousands of switches and links.
3) Systems are used at Google to continuously verify topology matches intent, detect routing inconsistencies within milliseconds, and measure service level agreements and traffic characteristics across all host pairs.
Cloud computing
The document provides an overview of cloud computing, including its history, models, architecture, security concerns, and importance. Cloud computing allows users to access software and store data on remote servers rather than local hardware. It has evolved from early concepts in the 1950s-60s to platforms like Amazon Web Services today. There are three main service models: Infrastructure as a Service, Platform as a Service, and Software as a Service. Cloud computing also raises issues around data protection, identity management, and compliance with varying legal requirements across jurisdictions. Overall it allows for improved data storage, maintenance, and security compared to traditional computing.
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
CQRS: Theory
This slide explains from from where CQRS comes from. It talks about distributed system and its fallacies and takes a look on CAP theorem.
A Low-Cost IoT Application for the Urban Traffic of Vehicles, Based on Wirele...
This document describes a low-cost IoT application to monitor urban traffic using wireless sensors and GSM technology. The system aims to address issues like traffic congestion, accidents, pollution and fuel consumption. It implements a distributed multilayer model using Arduino boards, laser sensors and a GSM module. Real-time traffic data is collected and stored in a database. Data mining techniques are then applied using the Pentaho platform to analyze traffic trends and patterns under different conditions. Functional and unit testing was performed to validate the system. The results from the deployment provide valuable traffic insights to help address congestion problems.
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...Agile Testing Alliance
This document discusses performance evaluation strategies for multi-access edge computing (MEC). It introduces MEC and outlines its advantages and challenges, including unpredictable network traffic from user mobility. It presents a performance model called MECCA and describes deployment models for testing MEC applications in cloud and edge-cloud environments. The testing approach focuses on key metrics like response time and throughput as more clients are added. Potential use cases for MEC include autonomous vehicles, healthcare, manufacturing, and augmented reality.Location Tracking and Smooth Path Providing System
This document describes a location tracking and route planning system that was created to address issues with GPS reliability and provide accurate travel routes. It uses machine learning and data science algorithms like Dijkstra's algorithm and Kalman filtering. The system takes a source and destination address from the user, processes the coordinates using graph networks and algorithms to find multiple routes and the shortest route. Kalman filtering is then applied to smooth the shortest route. The optimized route is plotted on a map along with nearby amenities, weather data, and travel time. The system aims to help traffic management and provide accurate paths to reduce travel efforts. It was implemented using Python libraries and APIs to retrieve location data, weather, and map features.
Presentation_final.pdf
The document proposes a federated learning approach for decentralized traffic flow prediction to address privacy and latency issues. Federated learning allows multiple nodes to build a shared machine learning model without sharing local datasets. The authors describe federated learning and its advantages like data security, diversity, and efficiency. They discuss applications and related work using federated learning for traffic prediction. The proposed approach uses a federated averaging algorithm to train a model across decentralized nodes holding local traffic data, aiming to accurately predict traffic flow while preserving data privacy.
Big data processing systems research
This document summarizes several systems for big data processing that extend or improve upon the MapReduce programming model. It discusses systems for iterative processing like HaLoop, stream processing like Muppet, improving performance through caching and indexing like Incoop and HAIL, and automatic optimization of MapReduce programs like MANIMAL and SkewTune. The document also briefly introduces broader distributed data processing frameworks beyond MapReduce like Dryad, SCOPE, Spark, Nephele/PACTs, and the ASTERIX scalable data platform.
Data Con LA 2018 - Enabling real-time exploration and analytics at scale at H...
Enabling real-time exploration and analytics at scale to drive operational intelligence at Hulu by Indrasis Mondal, Director, Data Engineering and Data Products, Hulu
Data is one of most powerful assets for companies today and a key driver for innovation, product development and business efficiency. Operational intelligence allows modern organization to use that data asset in real-time to enable immediate insights to their business operations and allow rapid decision making for strategic advantage. In this presentation we will walk through the operational intelligence capabilities Hulu has built to process tens of millions of events per minute to enable fast exploration of data and real-time decision making .
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
In this study, a mobile cloud offloading system has been developed to decide that a process run on the cloud or on the mobile platform. A context-aware decision algorithm has been developed. The low performance and problem of battery consumption of mobile devices have been fundamental challenges on the mobile computing. To overcome this kind of challenges, recent advances towards mobile cloud computing propose a selective mobile-to-cloud offloading service by moving a mobile application from a slow mobile device to a fast server in the cloud during run time. Determine whether a process running on cloud or not is an important issue. Power consumption and time limits are vitally important for decision. In this study we used PowerTutor application which is a dynamic power measurement modelling tool. Another important factor is the process completion time. Calculate the power consumption is very difficult
Superset druid realtime
This document discusses using real-time data with Superset by ingesting data streams using Apache Flink and storing the data in Apache Druid. It outlines use cases like operational insights and service debugging. It proposes using Flink for its scalable streaming capabilities and SQL interface and integrating with Druid for its columnar storage and geospatial support. It also describes transforming the data through UDFs before querying and visualizing it in Superset.
The RECAP Project: Large Scale Simulation Framework
The RECAP project aims to develop an architecture for reliable capacity provisioning and enhanced remediation for distributed cloud/edge/fog applications. The architecture includes a feedback loop consisting of a collector, application modeler, workload modeler, optimizer, and simulator. The collector gathers metrics across the infrastructure. The modelers develop models of application behavior and workloads. The optimizer performs optimization tasks like application placement based on these models. The simulator assists the optimizer by simulating different deployments. The project will apply this architecture to use cases from partners in telecommunications, analytics, IoT, and infrastructure management.
Exploring the latest and greatest from Java 14
Miro Cupak presented at the J4K 2020 conference on exploring the latest features in Java 14, including new productivity language features like switch expressions and text blocks as well as new APIs like flight recorder and foreign memory access. He discussed upcoming features and then provided a live coding demonstration.
Exploring reactive programming in Java
Miro Cupak gave a presentation on exploring reactive programming in Java, discussing 9 levels of reactive programming supported in Java, from basic threading in Java 1+ to fully reactive microservices using libraries like Vert.x, Spring WebFlux, and Kafka. The presentation covered live coding examples and discussed the different levels from low-level threading to high-level reactive libraries and services.
More Related Content
Similar to Odessa Enabling Interactive Perception Applications on Mobile Devices
Cloud Migration Strategy and Best Practices
Analyze key aspects to be considered before embarking on your cloud journey. The presentation outlines the strategies, approach, and choices that need to be made, to ensure a smooth transition to the cloud.
cloud computing.pptx fundamentals and deployment models
Cloud computing allows sharing of data, storage, and access to computer resources over the internet. It has three main service models: Infrastructure as a Service (IaaS) provides basic services and hardware that users maintain; Platform as a Service (PaaS) provides software, hardware, and platforms automatically; and Software as a Service (SaaS) provides transparent access to software and hardware. Cloud computing also has four deployment models: public, community, private, and hybrid clouds. While cloud computing provides benefits, it also faces challenges regarding privacy, compliance with regulations, security, and potential abuse.
Presented by Ahmed Abdulhakim Al-Absi - Scaling map reduce applications acro...
Scaling map reduce applications across hybrid clouds to meet soft deadlines - By Michael Mattess, Rodrigo N. Calheiros, and Rajkumar Buyya, Proceedings of the 27th IEEE International Conference on Advanced Information Networking and Applications (AINA 2013, IEEE CS Press, USA), Barcelona, Spain, March 25-28, 2013.
RECAP Project Overview
This a RECAP project overview slide deck prepared by Thang Le Duc (UMU), P-O Östberg (UMU) and Tomas Brännström (Tieto). It starts with an introduction and continues with a section on challenges for a self-orchestrated, self-remediated cloud system. It then presents the RECAP vision and use cases and finishes with a conclusion.
Trafficshifting: Avoiding Disasters & Improving Performance at Scale
LinkedIn uses traffic shifting techniques to avoid disasters and improve performance. They shift traffic between their datacenter fabrics and point of presence (PoP) locations using sticky routing and DNS steering. This allows them to perform disaster recovery testing and mitigate outages. Specifically for the Asia Pacific region, they overcame challenges such as fiber cuts by failing traffic out of affected PoP locations. Their techniques improved performance and user experience for both IPv4 and IPv6 users.
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice by José I. Benedetto Andrés Neyem Jaime Navón Guillermo Valenzuela. MobileSoft 2017, Buenos Aires.
APRICOT 2017: Trafficshifting: Avoiding Disasters & Improving Performance at ...
LinkedIn serves traffic for its 467 million members from four data centers and multiple PoPs spread geographically around the world. Serving live traffic from from many places at the same time has taken us from a disaster recovery model to a disaster avoidance model where we can take an unhealthy data center or PoP out of rotation and redistribute its traffic to a healthy one within minutes, with virtually no visible impact to users. The geographical distribution of our infrastructure also allows us to optimize the end-user's experience by geo routing users to the best possible PoP and datacenter.
This talk provide details on how LinkedIn shifts traffic between its PoPs and data centers to provide the best possible performance and availability for its members. We will also touch on the complexities of performance in APAC, how IPv6 is helping our members and how LinkedIn stress tests data centers verify its disaster recovery capabilities.
5G-USA-Telemetry
1) Google has built one of the fastest and most capable network infrastructures over the past 15+ years through innovations like global caching, software defined networking, and virtualizing the physical network.
2) Telemetry and analytics are needed in large data center networks to perform network modeling, configuration verification, and fault isolation given their complexity with thousands of switches and links.
3) Systems are used at Google to continuously verify topology matches intent, detect routing inconsistencies within milliseconds, and measure service level agreements and traffic characteristics across all host pairs.
Cloud computing
The document provides an overview of cloud computing, including its history, models, architecture, security concerns, and importance. Cloud computing allows users to access software and store data on remote servers rather than local hardware. It has evolved from early concepts in the 1950s-60s to platforms like Amazon Web Services today. There are three main service models: Infrastructure as a Service, Platform as a Service, and Software as a Service. Cloud computing also raises issues around data protection, identity management, and compliance with varying legal requirements across jurisdictions. Overall it allows for improved data storage, maintenance, and security compared to traditional computing.
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
CQRS: Theory
This slide explains from from where CQRS comes from. It talks about distributed system and its fallacies and takes a look on CAP theorem.
A Low-Cost IoT Application for the Urban Traffic of Vehicles, Based on Wirele...
This document describes a low-cost IoT application to monitor urban traffic using wireless sensors and GSM technology. The system aims to address issues like traffic congestion, accidents, pollution and fuel consumption. It implements a distributed multilayer model using Arduino boards, laser sensors and a GSM module. Real-time traffic data is collected and stored in a database. Data mining techniques are then applied using the Pentaho platform to analyze traffic trends and patterns under different conditions. Functional and unit testing was performed to validate the system. The results from the deployment provide valuable traffic insights to help address congestion problems.
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...Agile Testing Alliance
This document discusses performance evaluation strategies for multi-access edge computing (MEC). It introduces MEC and outlines its advantages and challenges, including unpredictable network traffic from user mobility. It presents a performance model called MECCA and describes deployment models for testing MEC applications in cloud and edge-cloud environments. The testing approach focuses on key metrics like response time and throughput as more clients are added. Potential use cases for MEC include autonomous vehicles, healthcare, manufacturing, and augmented reality.Location Tracking and Smooth Path Providing System
This document describes a location tracking and route planning system that was created to address issues with GPS reliability and provide accurate travel routes. It uses machine learning and data science algorithms like Dijkstra's algorithm and Kalman filtering. The system takes a source and destination address from the user, processes the coordinates using graph networks and algorithms to find multiple routes and the shortest route. Kalman filtering is then applied to smooth the shortest route. The optimized route is plotted on a map along with nearby amenities, weather data, and travel time. The system aims to help traffic management and provide accurate paths to reduce travel efforts. It was implemented using Python libraries and APIs to retrieve location data, weather, and map features.
Presentation_final.pdf
The document proposes a federated learning approach for decentralized traffic flow prediction to address privacy and latency issues. Federated learning allows multiple nodes to build a shared machine learning model without sharing local datasets. The authors describe federated learning and its advantages like data security, diversity, and efficiency. They discuss applications and related work using federated learning for traffic prediction. The proposed approach uses a federated averaging algorithm to train a model across decentralized nodes holding local traffic data, aiming to accurately predict traffic flow while preserving data privacy.
Big data processing systems research
This document summarizes several systems for big data processing that extend or improve upon the MapReduce programming model. It discusses systems for iterative processing like HaLoop, stream processing like Muppet, improving performance through caching and indexing like Incoop and HAIL, and automatic optimization of MapReduce programs like MANIMAL and SkewTune. The document also briefly introduces broader distributed data processing frameworks beyond MapReduce like Dryad, SCOPE, Spark, Nephele/PACTs, and the ASTERIX scalable data platform.
Data Con LA 2018 - Enabling real-time exploration and analytics at scale at H...
Enabling real-time exploration and analytics at scale to drive operational intelligence at Hulu by Indrasis Mondal, Director, Data Engineering and Data Products, Hulu
Data is one of most powerful assets for companies today and a key driver for innovation, product development and business efficiency. Operational intelligence allows modern organization to use that data asset in real-time to enable immediate insights to their business operations and allow rapid decision making for strategic advantage. In this presentation we will walk through the operational intelligence capabilities Hulu has built to process tens of millions of events per minute to enable fast exploration of data and real-time decision making .
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
In this study, a mobile cloud offloading system has been developed to decide that a process run on the cloud or on the mobile platform. A context-aware decision algorithm has been developed. The low performance and problem of battery consumption of mobile devices have been fundamental challenges on the mobile computing. To overcome this kind of challenges, recent advances towards mobile cloud computing propose a selective mobile-to-cloud offloading service by moving a mobile application from a slow mobile device to a fast server in the cloud during run time. Determine whether a process running on cloud or not is an important issue. Power consumption and time limits are vitally important for decision. In this study we used PowerTutor application which is a dynamic power measurement modelling tool. Another important factor is the process completion time. Calculate the power consumption is very difficult
Superset druid realtime
This document discusses using real-time data with Superset by ingesting data streams using Apache Flink and storing the data in Apache Druid. It outlines use cases like operational insights and service debugging. It proposes using Flink for its scalable streaming capabilities and SQL interface and integrating with Druid for its columnar storage and geospatial support. It also describes transforming the data through UDFs before querying and visualizing it in Superset.
The RECAP Project: Large Scale Simulation Framework
The RECAP project aims to develop an architecture for reliable capacity provisioning and enhanced remediation for distributed cloud/edge/fog applications. The architecture includes a feedback loop consisting of a collector, application modeler, workload modeler, optimizer, and simulator. The collector gathers metrics across the infrastructure. The modelers develop models of application behavior and workloads. The optimizer performs optimization tasks like application placement based on these models. The simulator assists the optimizer by simulating different deployments. The project will apply this architecture to use cases from partners in telecommunications, analytics, IoT, and infrastructure management.
Similar to Odessa Enabling Interactive Perception Applications on Mobile Devices (20)
cloud computing.pptx fundamentals and deployment models
cloud computing.pptx fundamentals and deployment models
Presented by Ahmed Abdulhakim Al-Absi - Scaling map reduce applications acro...
Presented by Ahmed Abdulhakim Al-Absi - Scaling map reduce applications acro...
Trafficshifting: Avoiding Disasters & Improving Performance at Scale
Trafficshifting: Avoiding Disasters & Improving Performance at Scale
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice
Rethinking the Mobile Code Offloading Paradigm: From Concept to Practice
APRICOT 2017: Trafficshifting: Avoiding Disasters & Improving Performance at ...
APRICOT 2017: Trafficshifting: Avoiding Disasters & Improving Performance at ...
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
WSO2CON 2024 - Designing Event-Driven Enterprises: Stories of Transformation
A Low-Cost IoT Application for the Urban Traffic of Vehicles, Based on Wirele...
A Low-Cost IoT Application for the Urban Traffic of Vehicles, Based on Wirele...
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...
#ATAGTR2021 Presentation : "Performance Evaluation Strategy of multi-access e...
Location Tracking and Smooth Path Providing System
Location Tracking and Smooth Path Providing System
Data Con LA 2018 - Enabling real-time exploration and analytics at scale at H...
Data Con LA 2018 - Enabling real-time exploration and analytics at scale at H...
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
CONTEXT-AWARE DECISION MAKING SYSTEM FOR MOBILE CLOUD OFFLOADING
The RECAP Project: Large Scale Simulation Framework
The RECAP Project: Large Scale Simulation Framework
More from Miro Cupak
Exploring the latest and greatest from Java 14
Miro Cupak presented at the J4K 2020 conference on exploring the latest features in Java 14, including new productivity language features like switch expressions and text blocks as well as new APIs like flight recorder and foreign memory access. He discussed upcoming features and then provided a live coding demonstration.
Exploring reactive programming in Java
Miro Cupak gave a presentation on exploring reactive programming in Java, discussing 9 levels of reactive programming supported in Java, from basic threading in Java 1+ to fully reactive microservices using libraries like Vert.x, Spring WebFlux, and Kafka. The presentation covered live coding examples and discussed the different levels from low-level threading to high-level reactive libraries and services.
Exploring the last year of Java
This document discusses recent and upcoming features in Java, including single-file source programs, improved HTTP/2 client API, productivity features like switch expressions and pattern matching, text blocks, records, and a new flight recorder API. The author will provide a live demo and take questions on exploring the last year of Java development.
Local variable type inference - Will it compile?
This document discusses local variable type inference, asking whether code will compile without explicitly declaring variable types, and is authored by Miro Cupak, Co-founder and VP Engineering at DNAstack. The author invites questions on the topic of local variable type inference.
The Good, the Bad and the Ugly of Java API design
The document discusses Java API design patterns and antipatterns based on several Java APIs, including:
1. Collections API - Take advantage of immutable collections created with factory methods and copies to avoid issues with shared references.
2. Stack-walking API - Leverage the Stream API to lazily access call stack elements and avoid manually resolving classes.
3. Process API - Use ProcessHandle for a clean way to obtain process information with convenience methods rather than external libraries.
4. HTTP client API - Prefer the new HTTP client over HttpURLConnection for a clear, versatile and flexible design with asynchronous support.
Local variable type inference - Will it compile?
Local variable type inference in Java allows the compiler to infer the type of local variables declared with "var" instead of explicitly stating the type. While it reduces boilerplate and improves readability, it has some limitations and best practices should be followed when using it. Specifically, var cannot be used for declarations without initializers, null/array initializers, compound declarations, method signatures, class fields, catch formals, lambdas or method references. Developer needs to consider whether the inferred type will be obvious, be careful of primitive types being inferred as int, and avoid overuse with generics.
Exploring reactive programming in Java
Miro Cupak explores 8 levels of reactive programming in Java, from the basic Thread and Runnable interfaces introduced in Java 1 to higher levels including reactive streams in Java 9, the HTTP/2 client in Java 11, and reactive libraries and microservices. The tweet thread outlines each level, noting the Java version it was introduced in, with the goal of explaining reactive programming concepts and discussing any questions.
The good, the bad, and the ugly of Java API design
Miro Cupak discusses the good and bad aspects of several Java API designs, including Collections, Stack-walking, Process, and HTTP client APIs. Some best practices highlighted are convenience factory methods, immutability by default, lazy evaluation using streams, and clear separation of concepts. Antipatterns include unnecessary dependencies, unexpected behavior, and side effects. Well-designed APIs use static methods, fluent interfaces, and asynchronous programming with CompletableFuture.
Master class in modern Java
This document contains the agenda and tasks for a workshop on modern Java features. The workshop will be held on October 30, 2019 from 9:15am to 6pm with a lunch break from 1:15pm to 2pm. It will cover 10 lessons including JShell, collections factory methods, try-with-resources improvements, Stream API enhancements, Optional extensions, CompletableFuture updates, and reactive streams. Each lesson contains multiple tasks for participants to complete to explore the featured Java topics hands-on.
The good, the bad, and the ugly of Java API design
Slides for my The good, the bad, and the ugly of Java API design talk at Detroit Java User Group 2019.
Exploring reactive programming in Java
Miro Cupak gave a presentation on exploring reactive programming in Java. He discussed 8 levels of reactivity in Java, ranging from basic threading concepts in early versions of Java to fully reactive microservices using libraries like Vert.x and frameworks like Spring WebFlux. The talk included a live demo and concluded with an opportunity for questions.
The good, the bad, and the ugly of Java API design
The document discusses various Java API design patterns and antipatterns related to collections, stack walking, process handling, and HTTP clients. Some best practices highlighted include using immutable collections by default, leveraging streams, providing convenience methods, and having a clear separation of concepts. Antipatterns include unnecessary dependencies, unexpected behavior, and inconsistent naming. The document advocates preferring functionality in the JDK over external libraries.
Writing clean code with modern Java
This document summarizes new features in modern Java including private methods in interfaces, improved try-with-resources, switch expressions, and local variable type inference. It provides examples and best practices for each feature, such as using private methods to keep default methods short and clear, always preferring try-with-resources over try-finally, and being careful when combining local variable type inference with generics. The document aims to help developers write cleaner code with the new Java features.
The good, the bad, and the ugly of Java API design
The document discusses the good and bad aspects of Java API design based on the Collections API, Stack-walking API, Process API, and HTTP client API. Some best practices highlighted include using immutable collections by default, static factory methods to create objects, and convenience methods. Antipatterns include obtaining collections through other data structures and manually resolving classes. Patterns referenced are lazy access via streams and providing adapters between related types.
Master class in modern Java
This document contains the schedule and lessons for a workshop on modern Java features. The schedule includes registration, two workshop sessions with a lunch break in between. The lessons cover JShell, collection factory methods, try-with-resources improvements, Stream API enhancements, Optional extensions, CompletableFuture updates, reactive streams, and more. The document provides tasks for participants to complete to learn each lesson.
Exploring reactive programming in Java
Miro Cupak presented on exploring reactive programming in Java, demonstrating different levels of reactivity available in Java from threads and runnables to reactive streams and libraries to fully reactive microservices. The presentation covered 8 levels of reactive programming capabilities in Java from the earliest versions to the newest, including executors, fork/join pools, completable futures, reactive streams, HTTP/2 clients, and reactive libraries and frameworks. Cupak concluded by taking questions on Twitter.
Writing clean code with modern Java
This document discusses writing clean code with modern Java features such as immutable collections, private methods in interfaces, switch expressions, and local variable type inference. It provides tips for using each feature, such as preferring immutable collections by default and only adding mutability when needed, using private methods in interfaces to keep default methods short, and using local variable type inference carefully and only for variables with explicit initializers. The document encourages writing clean, readable code with these new Java language features.
Exploring what's new in Java 10 and 11 (and 12)
This document is a presentation on new features in recent and upcoming versions of Java, including Java 10, 11, and 12. It includes a live demo section and leaves time at the end for questions. The session notes will be shared on Twitter.
Exploring what's new in Java 10 and 11
This document is a summary of a presentation on new features in Java 10 and 11. It includes the presenter's Twitter handle, an invitation to a live demo during the presentation, and a request for questions from the audience with an assigned hashtag. The session notes from the presentation will be shared on Twitter.
Exploring what's new in Java in 2018
This Twitter thread from Miro Cupak discusses what's new in Java in 2018. It includes a live demo as well as an invitation for questions from attendees. Additional session notes were provided on Twitter.
More from Miro Cupak (20)
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
The good, the bad, and the ugly of Java API design
Recently uploaded
What’s New in Odoo 17 – A Complete Roadmap
Odoo releases a new update every year. The latest version, Odoo 17, came out in October 2023. It brought many improvements to the user interface and user experience, along with new features in modules like accounting, marketing, manufacturing, websites, and more.
The Odoo 17 update has been a hot topic among startups, mid-sized businesses, large enterprises, and Odoo developers aiming to grow their businesses. Since it is now already the first quarter of 2024, you must have a clear idea of what Odoo 17 entails and what it can offer your business if you are still not aware of it.
This blog covers the features and functionalities. Explore the entire blog and get in touch with expert Odoo ERP consultants to leverage Odoo 17 and its features for your business too.
An Overview of Odoo ERP
Odoo ERP was first released as OpenERP software in February 2005. It is a suite of business applications used for ERP, CRM, eCommerce, websites, and project management. Ten years ago, the Odoo Enterprise edition was launched to help fund the Odoo Community version.
When you compare Odoo Community and Enterprise, the Enterprise edition offers exclusive features like mobile app access, Odoo Studio customisation, Odoo hosting, and unlimited functional support.
Today, Odoo is a well-known name used by companies of all sizes across various industries, including manufacturing, retail, accounting, marketing, healthcare, IT consulting, and R&D.
The latest version, Odoo 17, has been available since October 2023. Key highlights of this update include:
Enhanced user experience with improvements to the command bar, faster backend page loading, and multiple dashboard views.
Instant report generation, credit limit alerts for sales and invoices, separate OCR settings for invoice creation, and an auto-complete feature for forms in the accounting module.
Improved image handling and global attribute changes for mailing lists in email marketing.
A default auto-signature option and a refuse-to-sign option in HR modules.
Options to divide and merge manufacturing orders, track the status of manufacturing orders, and more in the MRP module.
Dark mode in Odoo 17.
Now that the Odoo 17 announcement is official, let’s look at what’s new in Odoo 17!
What is Odoo ERP 17?
Odoo 17 is the latest version of one of the world’s leading open-source enterprise ERPs. This version has come up with significant improvements explained here in this blog. Also, this new version aims to introduce features that enhance time-saving, efficiency, and productivity for users across various organisations.
Odoo 17, released at the Odoo Experience 2023, brought notable improvements to the user interface and added new functionalities with enhancements in performance, accessibility, data analysis, and management, further expanding its reach in the market.
Baha Majid WCA4Z IBM Z Customer Council Boston June 2024.pdf
IBM watsonx Code Assistant for Z, our latest Generative AI-assisted mainframe application modernization solution. Mainframe (IBM Z) application modernization is a topic that every mainframe client is addressing to various degrees today, driven largely from digital transformation. With generative AI comes the opportunity to reimagine the mainframe application modernization experience. Infusing generative AI will enable speed and trust, help de-risk, and lower total costs associated with heavy-lifting application modernization initiatives. This document provides an overview of the IBM watsonx Code Assistant for Z which uses the power of generative AI to make it easier for developers to selectively modernize COBOL business services while maintaining mainframe qualities of service.
Using Query Store in Azure PostgreSQL to Understand Query Performance
Microsoft has added an excellent new extension in PostgreSQL on their Azure Platform. This session, presented at Posette 2024, covers what Query Store is and the types of information you can get out of it.
Mobile App Development Company In Noida | Drona Infotech
React.js, a JavaScript library developed by Facebook, has gained immense popularity for building user interfaces, especially for single-page applications. Over the years, React has evolved and expanded its capabilities, becoming a preferred choice for mobile app development. This article will explore why React.js is an excellent choice for the Best Mobile App development company in Noida.
Visit Us For Information: https://www.linkedin.com/pulse/what-makes-reactjs-stand-out-mobile-app-development-rajesh-rai-pihvf/
A Comprehensive Guide on Implementing Real-World Mobile Testing Strategies fo...
In today's fiercely competitive mobile app market, the role of the QA team is pivotal for continuous improvement and sustained success. Effective testing strategies are essential to navigate the challenges confidently and precisely. Ensuring the perfection of mobile apps before they reach end-users requires thoughtful decisions in the testing plan.
ppt on the brain chip neuralink.pptx
A neural network is a machine learning program, or model, that makes decisions in a manner similar to the human brain, by using processes that mimic the way biological neurons work together to identify phenomena, weigh options and arrive at conclusions.
Enums On Steroids - let's look at sealed classes !
These are slides for my session"Enums On Steroids - let's look at sealed classes !" - delivered among others, on Devoxx UK 2024 conference
Malibou Pitch Deck For Its €3M Seed Round
French start-up Malibou raised a €3 million Seed Round to develop its payroll and human resources
management platform for VSEs and SMEs. The financing round was led by investors Breega, Y Combinator, and FCVC.
Transforming Product Development using OnePlan To Boost Efficiency and Innova...
Ready to overcome challenges and drive innovation in your organization? Join us in our upcoming webinar where we discuss how to combat resource limitations, scope creep, and the difficulties of aligning your projects with strategic goals. Discover how OnePlan can revolutionize your product development processes, helping your team to innovate faster, manage resources more effectively, and deliver exceptional results.
J-Spring 2024 - Going serverless with Quarkus, GraalVM native images and AWS ...
Presented at NLJUG's J-Spring 2024.
Alluxio Webinar | 10x Faster Trino Queries on Your Data Platform
Alluxio Webinar
June. 18, 2024
For more Alluxio Events: https://www.alluxio.io/events/
Speaker:
- Jianjian Xie (Staff Software Engineer, Alluxio)
As Trino users increasingly rely on cloud object storage for retrieving data, speed and cloud cost have become major challenges. The separation of compute and storage creates latency challenges when querying datasets; scanning data between storage and compute tiers becomes I/O bound. On the other hand, cloud API costs related to GET/LIST operations and cross-region data transfer add up quickly.
The newly introduced Trino file system cache by Alluxio aims to overcome the above challenges. In this session, Jianjian will dive into Trino data caching strategies, the latest test results, and discuss the multi-level caching architecture. This architecture makes Trino 10x faster for data lakes of any scale, from GB to EB.
What you will learn:
- Challenges relating to the speed and costs of running Trino in the cloud
- The new Trino file system cache feature overview, including the latest development status and test results
- A multi-level cache framework for maximized speed, including Trino file system cache and Alluxio distributed cache
- Real-world cases, including a large online payment firm and a top ridesharing company
- The future roadmap of Trino file system cache and Trino-Alluxio integration
DevOps Consulting Company | Hire DevOps Services
Spiral Mantra excels in providing comprehensive DevOps services, including Azure and AWS DevOps solutions. As a top DevOps consulting company, we offer controlled services, cloud DevOps, and expert consulting nationwide, including Houston and New York. Our skilled DevOps engineers ensure seamless integration and optimized operations for your business. Choose Spiral Mantra for superior DevOps services.
https://www.spiralmantra.com/devops/
8 Best Automated Android App Testing Tool and Framework in 2024.pdf
Regarding mobile operating systems, two major players dominate our thoughts: Android and iPhone. With Android leading the market, software development companies are focused on delivering apps compatible with this OS. Ensuring an app's functionality across various Android devices, OS versions, and hardware specifications is critical, making Android app testing essential.
Operational ease MuleSoft and Salesforce Service Cloud Solution v1.0.pptx
Operational ease MuleSoft and Salesforce Service Cloud Solution v1.0
Manyata Tech Park Bangalore_ Infrastructure, Facilities and More
Located in the bustling city of Bangalore, Manyata Tech Park stands as one of India’s largest and most prominent tech parks, playing a pivotal role in shaping the city’s reputation as the Silicon Valley of India. Established to cater to the burgeoning IT and technology sectors
Unlock the Secrets to Effortless Video Creation with Invideo: Your Ultimate G...
Unlock the Secrets to Effortless Video Creation with Invideo: Your Ultimate G...The Third Creative Media
"Navigating Invideo: A Comprehensive Guide" is an essential resource for anyone looking to master Invideo, an AI-powered video creation tool. This guide provides step-by-step instructions, helpful tips, and comparisons with other AI video creators. Whether you're a beginner or an experienced video editor, you'll find valuable insights to enhance your video projects and bring your creative ideas to life.How Can Hiring A Mobile App Development Company Help Your Business Grow?
ToXSL Technologies is an award-winning Mobile App Development Company in Dubai that helps businesses reshape their digital possibilities with custom app services. As a top app development company in Dubai, we offer highly engaging iOS & Android app solutions. https://rb.gy/necdnt
Why Apache Kafka Clusters Are Like Galaxies (And Other Cosmic Kafka Quandarie...
Closing talk for the Performance Engineering track at Community Over Code EU (Bratislava, Slovakia, June 5 2024) https://eu.communityovercode.org/sessions/2024/why-apache-kafka-clusters-are-like-galaxies-and-other-cosmic-kafka-quandaries-explored/ Instaclustr (now part of NetApp) manages 100s of Apache Kafka clusters of many different sizes, for a variety of use cases and customers. For the last 7 years I’ve been focused outwardly on exploring Kafka application development challenges, but recently I decided to look inward and see what I could discover about the performance, scalability and resource characteristics of the Kafka clusters themselves. Using a suite of Performance Engineering techniques, I will reveal some surprising discoveries about cosmic Kafka mysteries in our data centres, related to: cluster sizes and distribution (using Zipf’s Law), horizontal vs. vertical scalability, and predicting Kafka performance using metrics, modelling and regression techniques. These insights are relevant to Kafka developers and operators.
TMU毕业证书精仿办理
多伦多都会大学毕业证成绩单(留信网认证)【176555708微信】不能毕业办理假文凭回国找工作
【多伦多都会大学毕业证成绩单仿制】（176555708微信☀️）伪造研究生学位证书TMU毕业证文凭,多伦多都会大学文凭证书学籍信息入库,多伦多都会大学学位证书留学高校学位证书补办
【TMU毕业证【微信176555708】多伦多都会大学文凭学历】【微信176555708】【多伦多都会大学文凭学历证书】【多伦多都会大学毕业证书与成绩单样本图片】毕业证书补办 Fake Degree做学费单【毕业证明信-推荐信】成绩单，录取通知书，Offer，在读证明，雅思托福成绩单，真实大使馆教育部认证，回国人员证明，留信网认证。网上存档永久可查！
Recently uploaded (20)
Baha Majid WCA4Z IBM Z Customer Council Boston June 2024.pdf
Baha Majid WCA4Z IBM Z Customer Council Boston June 2024.pdf
Using Query Store in Azure PostgreSQL to Understand Query Performance
Using Query Store in Azure PostgreSQL to Understand Query Performance
Mobile App Development Company In Noida | Drona Infotech
Mobile App Development Company In Noida | Drona Infotech
A Comprehensive Guide on Implementing Real-World Mobile Testing Strategies fo...
A Comprehensive Guide on Implementing Real-World Mobile Testing Strategies fo...
Enums On Steroids - let's look at sealed classes !
Enums On Steroids - let's look at sealed classes !
Transforming Product Development using OnePlan To Boost Efficiency and Innova...
Transforming Product Development using OnePlan To Boost Efficiency and Innova...
J-Spring 2024 - Going serverless with Quarkus, GraalVM native images and AWS ...
J-Spring 2024 - Going serverless with Quarkus, GraalVM native images and AWS ...
Alluxio Webinar | 10x Faster Trino Queries on Your Data Platform
Alluxio Webinar | 10x Faster Trino Queries on Your Data Platform
8 Best Automated Android App Testing Tool and Framework in 2024.pdf
8 Best Automated Android App Testing Tool and Framework in 2024.pdf
Operational ease MuleSoft and Salesforce Service Cloud Solution v1.0.pptx
Operational ease MuleSoft and Salesforce Service Cloud Solution v1.0.pptx
Manyata Tech Park Bangalore_ Infrastructure, Facilities and More
Manyata Tech Park Bangalore_ Infrastructure, Facilities and More
Unlock the Secrets to Effortless Video Creation with Invideo: Your Ultimate G...
Unlock the Secrets to Effortless Video Creation with Invideo: Your Ultimate G...
How Can Hiring A Mobile App Development Company Help Your Business Grow?
How Can Hiring A Mobile App Development Company Help Your Business Grow?
Why Apache Kafka Clusters Are Like Galaxies (And Other Cosmic Kafka Quandarie...
Why Apache Kafka Clusters Are Like Galaxies (And Other Cosmic Kafka Quandarie...
Odessa Enabling Interactive Perception Applications on Mobile Devices
- 1. Odessa Enabling Interactive Perception Applications on Mobile Devices Miroslav Cupák Moo-Ryong Ra, Anmol Sheth, Lily Mummertx, Padmanabhan Pillai, David Wetherallo, Ramesh Govindan 25/03/2013 Some slides borrowed from http://catnip.usc.edu/home/.
- 2. Outline ● Problem & Contributions ● Metrics ● Applications ● Sprout ● Application Performance ● Design & Implementation ● Evaluation ● Conclusions
- 3. Problem ● resource constrained mobile devices ● offloading and parallelism ● interactive perception applications ● requirements ○ crisp response ○ continuous processing ○ compute-intensive ○ performance depending on data
- 4. Contributions ● study of offloading & parallelism decisions ● Odessa ○ runtime automatically determining how best to use offloading and parallelism to achieve responsiveness and accuracy ● successful evaluation of the greedy approach
- 5. Metrics ● makespan ○ time to execute all stages of computation for a frame ● throughput ○ rate at which frames are processed ● techniques ○ code offloading ○ pipelining ○ data parallelism
- 6. Applications
- 7. Sprout ● distributed stream processing system ● uses a data flow model to dynamically distribute processing of an application ● takes care of data transfer, coarse-grained data parallelism, pipeline parallelism ● Odessa ○ what/when/how to offload on top of Sprout
- 8. Application Performance ● observations ○ offloading must be made dynamically (input variability, network bandwidth, device parameters) ○ data parallelism cannot be determined apriori ○ static choice of pipeline parallelism is not optimal ● setup ○ netbook (1 core), laptop (2 cores), server (8 cores) ○ 30 fps, 640x480 px video, 3000 (500) frames ○ network bandwidth emulation
- 10. Variability Across Mobile Platforms
- 12. Data & Pipeline Parallelism
- 13. Design ● goals ○ minimize makespan and maximize throughput ○ quick response to environment changes ○ low computation and communication overhead ● parts ○ application profiler ○ decision engine
- 14. Profiler ● lightweight, online ● piggybacking approach ● function ○ maintain application performance profile for the decision engine ● data ○ execution time of each processing stage ○ wait time on connectors ○ transfer time
- 15. Decision Engine ● runs periodically ● greedy incremental approach ● estimates impact of offloading and parallelism based on processor frequencies and history of network measurements ● dynamic pipelining based on tokens
- 16. Decision Engine
- 17. Evaluation: Performance & Overhead ● minimum overhead
- 18. Evaluation: Other Strategies ● 3 strategies : offload all, domain-specific, offline optimizer
- 19. Evaluation: Context Adaptation ● taking fidelity into account left for future work
- 20. Related Work ● energy saving: MAUI ● graph-based partitioning: Coign ● static partitioning: Wishbone ● a set of pre-specified partitions: CloneCloud, Chroma, Spectra ● parallel processing: MapReduce
- 21. Conclusions ● understanding of the factors which contribute to offloading and parallelism decisions ● Odessa ○ runtime automatically determining how best to use offloading and parallelism to achieve responsiveness and accuracy ● extensive evaluation
- 23. Discussion ● Could static analysis be used to improve the approach? ● The paper focuses on computer vision problems. Would this approach work with other problems as well? ● Could this approach be altered to minimize energy consumption instead? ● How could the implementation be improved?