embedded systems cloud computing implementations - challenges

8/13/2019 Embedded systems cloud computing implementations - Challenges

1/4

Technical Insights - Embedded

FossilShale Embedded Technologies Confidential Page 1 of 4

Embedded Systems Cloud computing Implemetation - Challenges

The Embedded Technology industry is experiencing two major trends. On one hand, computation ismoving away from traditional desktop and department-level computer centers towards aninfrastructural core that consists of many large and distributed data centers with high-performancecomputer servers and data storage devices virtualized and available as Cloud services. These large-scalecenters provide all sorts of computational services to a multiplicity of peripheral clients, through variousinterconnection networks.

On the other hand, the increasing majority of these clients consist of a growing variety of embeddeddevices, such as smart phones, tablet computers and television set-top boxes (STB), whose capabilitiescontinue to improve while also providing data locality associated to data-intensive applicationprocessing of interest Indeed, the massive scale of todays data creation explosion is closely aligned tothe distributed computational resources of the expanding universe of distributed embedded systemsand devices. Multiple Service Operators (MSOs), such as cable providers, are an example of companiesthat drive both the rapid growth and evolution of large-scale computational systems, consumer andbusiness data, as well as the deployment of an increasing number of increasingly-powerful embeddedprocessors.


2/4



The distributed real-time and embedded (DRE) systems manage data and resources that are critical tothe ongoing system operations. Examples include testing and training of experimental air-craft across alarge geographic area, air traffic management systems, and disaster recovery operations. These types ofenterprise DRE systems must be configured correctly to leverage available resources and respond to thesystem deployment environment. For example, search and rescue missions in disaster recovery

operations need to configure the image resolution used to detect and track survivors depending on theavailable resources (e.g., computing power and network bandwidth)

Embedded Cloud Computing systems are implemented and developed for a specificcomputing/networking platform and deployed with the expectation of specific computing andnetworking resources being available at runtime. This approach simplifies development complexitysince system developers need only focus on how the system behaves in one operating environment,thereby ameliorating considerations of multiple infrastructure platforms with respect to systemquality-of-service (QoS) properties (e.g., responsiveness of computing platform, latency and reliabilityof networked data, etc.). Focusing on only a single operating environment, however, decreases theflexibility of the system and makes it hard to integrate into different operating environments, e.g.,porting to new computing and networking hardware.

Cloud computing is an increasingly popular infrastructure paradigm where computing and networkingresources are provided to a system or application as a service typically for a pay-as-you-go usagefee. Provisioning services in cloud environments relieve enterprise operators of many tedious tasksassociated with managing hardware and software resources used by systems and applications. Cloudcomputing also provides enterprise application developers and operators with additional flexibility byvirtualizing resources, such as providing virtual machines that can differ from the actual hardware

machines used.

Several middleware platforms such as the Java Message Service and Web Services BrokeredNotification can be used to the following

(1) Leverage cloud environments(2) Support large-scale data-centric distributed systems(3) Ease development and deployment of these systems.

These sub platforms, however, do not support fine-grained and robust QoS that are needed for DREsystems. Some large-scale distributed system platforms, such as the Global Information Grid andNetwork-centric Enterprise Services require rapid response, reliability, bandwidth guarantees,scalability, and fault-tolerance.

Conversely, conventional cloud environments are problematic for DRE systems since applications withinthese systems often cannot characterize the utilization of their specific resources (e.g., CPU speeds andmemory) accurately a priori. Consequently, applications in DRE systems may need to adjust to theavailable resources supplied by the cloud environment (e.g., using compression algorithms optimized for


3/4



given CPU power and memory) since the presence/absence of these resources affect timeliness andother QoS properties crucial to proper operation. If these adjustments take too long the mission that theDRE system supports could be jeopardized

Configuring an enterprise DRE pub/sub system in a cloud environment is hard be-cause the DRE system

must understand how the computing and networking resources affect end-to-end QoS. For example,transport protocols provide different types of QoS (e.g., reliability and latency) that must be configuredin conjunction with the pub/sub middleware. To work properly, however, QoS-enabled pub/submiddleware must un-derstand how these protocols behave with different cloud infrastructures.Likewise, the middleware must be configured with appropriate transport protocols to support the re-quired end-to-end QoS. Manual or ad hoc configuration of the transport and middleware can be tedious,error-prone, and time consuming.

Challenges in Embedded Cloud Computing Middleware

Challenge 1: Configuring for data timeliness and reliability.

Embedded Cloud Computing operations must receive sufficient data reliability and timeliness so thatmultiple data streams can be fused appropriately.

For instance, the data streams (such as infrared scan and video streams) can be exploited by multipleapplications simultaneously in a datacenter. Security monitoring and structural damage applications canuse video stream data to detect looting and unsafe buildings, respectively. The fire detectionapplications and power grid assessment applications can use infrared scans to detect fires and workingHVAC systems, respectively.

Embedded Cloud Computing systems must be configured to best use the computing and net-workingresources from the cloud to address data timeliness and reliability. These systems must therefore (1)use transport protocols that provide both reliability and timeliness and (2) know how these protocolsbehave in different computing and networking environments

Challenge 2: Timely configuration.

Due to timeliness concerns of Embedded Cloud Computing systems such as remotely controlledsystems, the ad hoc datacenter used for remotely controlled and monitor operations must beconfigured in a timely manner based on the computing and networking resources provided by the cloud.If the datacenter cannot be configured quickly, invaluable time will be lost leading to survivors not beingsaved and critical infrastructure (such as dams and power plants) not being safeguarded from furtherdamage. During a regional or emergency any wasted time can mean the difference between life anddeath for the systems and the salvaging or destruction of key regional utilities.

Moreover, applications and systems used during one of the critical functionality can be leveraged forother critical operations. Available computing and networking resources differ from one set of


4/4



operations to another. Depending on the available cloud resources, therefore, the configuration timesof e.g. , ad hoc datacenters for Embedded Cloud Computing systems operations must be bounded andfast to ensure appropriate responsiveness. Determining appropriate con-figurations must also providepredictable response to ensure rapid and dependable response times across different computing andnetworking resources.

Challenge 3: Accuracy of configurations.

Since data timeliness and reliability is related to the computing resources available and theconfiguration of the datacenter supporting the Embedded Cloud Computing system operations in acloud as noted in Challenge 1, configuring the datacenter must be done in an accurate manner. If thedatacenter is incorrectly configured then the timeliness and reliability of the data ( e.g. , the UAV scansand camera video used to detect survivors) will not be optimal for the given computing resources. Forcritical operations during Embedded Cloud Computing such as rescuing Embedded Cloud Computing systems system must utilize the available resources to their fullest extent.

Challenge 4: Reducing development complexity.

Embedded Cloud Computing system operations apply in many places and at many different applications.The functionality of applications used during Embedded Cloud Computing system operations may alsobe needed for other applications. A system that is developed for one particular application in aparticular operating environment, however, might not work well for a different application in a differentoperating environment. Embedded Cloud Computing system operations could unexpectedly fail at atime when they are needed most due to differences in computing and networking resources available.Systems therefore must be developed and configured readily between the different operatingenvironments presented by cloud computing to leverage the systems across a wide range of disasterscenarios.

embedded systems cloud computing implementations - challenges

Documents