This book describes the state-of-the-art in energy efficient, fault-tolerant embedded systems. It covers the entire product lifecycle of electronic systems design, analysis and testing and includes discussion of both circuit and system-level approaches. Readers will be enabled to meet the conflicting design objectives of energy efficiency and fault-tolerance for reliability, given the up-to-date techniques presented.

### Recommended for you

Springer Professional. Back to the search result list. Table of Contents Frontmatter Chapter 1. Embedded systems are making their way into more and more devices, from hand-held gadgets to household appliances, and from mobile devices to cars. The current trend is that this growth will continue and the market is expected to experience a three-fold rise in the demand from to [20].

From commercial to life-critical applications, the proliferation of computing systems in everyday life has substantially increased our dependence on them. Failures in air traffic control systems, nuclear reactors, or hospital patient monitoring systems can bring catastrophic consequences.

In order to enhance the dependability of computing systems, an effective evaluation of their reliability is desired. This chapter presents methods for evaluating system reliability, and indicates that stochastic modeling has provided an effective and unified framework for analyzing various aspects of reliability. While the relentless scaling of CMOS technology has brought digital IC designs with enhanced functionality and improved performance in every new generation, at the same time, the associated ever-increasing on-chip power and temperature densities make them suffer from more severe reliability threats [6, 93].

Digital electronic circuits are subject to many types of error. Considering the effect of such errors on the circuit functionality, they can be classed as permanent, transient or intermittent. Online since:. July Add to Cart. Cited by. Related Articles. Paper Title Pages. The instrument can automatically acquire data, the process of acquire data is convenient, fast, safe and the data is reliable. Abstract: The multiplexer act as the inputs of analog signals, enabling multi-channel analog signal acquisition; then filtered through a certain signal processing to ensure the most effective analog signals into the ADC; after conversion, the digital output is saved in the temporary buffer zone of the FPGA.

The whole process is controlled by the FPGA. Authors: Yi Wang Wang.

Abstract: A novel fan filter unit FFU motors group control system based on ZigBee wireless sensor networks is designed and implemented. The overall structure and composition of the proposed system were described. The group system communicate by the use of ZigBee wireless sensor network based CC,which effectively overcome the traditional FFU fan group control network wiring complexity and high cost disadvantages, and improve the flexibility and control accuracy, meanwhile reducing system cost.

The testing results show that the proposed system is feasible, accurate monitoring and control effectively, has the high valuable in engineering and marketing. The key problems of the system design are measurement method of automatic transmission solenoid duty, signal conditioning circuit design and multi-signal acquisition synchronization. The automatic transmission shift control strategy and oil pressure characteristics of clutch or brake etc can be analyzed through test data acquisition.

Abstract: As the unpredictability of market needs and the mass customization trends increase, employing reconfigurable industrial robotic work cells becomes a viable solution for manufacturers. A reconfigurable manufacturing system is a system designed for a quick change in structure, both in hardware and software, in order to rapidly adjust production capacity and functionalities. However, the approach is quite general; a simple search within scientific literature, patent databases or the world-wide-web returns thousands of results, covering a time period from the 80s to date.

Among the results, different approaches on reconfigurable industrial robotic work cell implementations can be easily noticed. More specifically, ing remote servers, energy consumption for accessing we investigate how to store data as well as process the them must be minimized while taking into account stored data in mobile cloud with k-out-of-n reliability the dynamically changing topology. As long as k or Manuscript received 29 Aug.

## Design of FPGA-Based Fault-Tolerant Embedded System

Similarly, another set of n nodes c IEEE. The param- eters k and n determine the degree of reliability and different k, n pairs may be assigned to data storage Application data3 func3 data2 func2 and data processing. System administrators select these data1 func1 parameters based on their reliability requirements. In order to process the data, applications duces the architecture of the framework and the math- provide functions that take the stored data as inputs.

Section 3 describes Each function is instantiated as multiple tasks that pro- the functions and implementation details of each com- cess the data simultaneously on different nodes. Nodes ponent in the framework. In section 4, an application executing tasks are processor nodes; we call a set of that uses our framework i. Client nodes system — MDFS is developed and evaluated. Section are the nodes requesting data allocation or processing 5 presents the performance evaluation of our k-out-of- operations.

A node can have any combination of roles n framework through extensive simulations. Section 6 from: storage node, processor node, or client node, and reviews the state of art. We conclude in Section 7.

- REINCARNATION MYSTERY SOLVED!
- 9781461441922 - Energy-Efficient Fault-Tolerant Systems (Embedded Systems) by mathews.
- The Architect.
- Clan Fraser, Once Removed.
- HMO Property Success: The proven strategy for financial freedom through multi-let property investing.
- Harriet Beamer Takes the Bus (Harriet Beamer Series).

As shown in Figure 1, our framework consists of 2 Architecture and Formulations five components: Topology Discovery and Monitoring, An overview of our proposed framework is depicted Failure Probability Estimation, Expected Transmission in Figure 1. The framework, running on all mobile Time ETT Computation, k-out-of-n Data Allocation nodes, provides services to applications that aim to: and k-out-of-n Data Processing.

When a request for 1 store data in mobile cloud reliably such that the data allocation or processing is received from applica- energy consumption for retrieving the data is minimized tions, the Topology Discovery and Monitoring compo- k-out-of-n data allocation problem ; and 2 reliably nent provides network topology information and fail- process the stored data such that energy consumption ure probabilities of nodes.

The failure probability is for processing the data is minimized k-out-of-n data estimated by the Failure Probability component on processing problem. As an example, an application each node. Based on the retrieved failure probabilities running in a mobile ad-hoc network may generate a and network topology, the ETT Computation com- large amount of media files and these files must be ponent computes the ETT matrix, which represents stored reliably such that they are recoverable even if the expected energy consumption for communication certain nodes fail.

At later time, the application may between any pair of node. Given the ETT matrix, our make queries to files for information such as the number framework finds the locations for storing fragments or of times an object appears in a set of images. Without executing tasks. The k-out-of-n Data Storage compo- loss of generality, we assume a data object is stored nent partitions data into n fragments by an erasure code once, but will be retrieved or accessed for processing algorithm and stores these fragments in the network multiple times later.

As shown in Figure 1, fragments by any node is minimized. If an data in the network. For higher data reliability and application needs to process the data, the k-out-of-n availability, each data is encoded and partitioned into Data Processing component creates a job of M tasks c IEEE. This component ensures that minimized. We formulate this problem as an ILP in all tasks complete as long as k or more processor nodes Equations 1 - 5. For convenience, we will use i and vi interchangeably hereafter.

Each node has that each node has access to k storage nodes; the an associated failure probability P [fi ] where fi is the third constraint Eq 4 ensures that j th column of event that causes node vi to fail. More decision variables. Storage node list X is a binary vector containing storage nodes, i. The ETT metric [7] has been widely used to retrieve and decode k data fragments because nodes for estimating transmission time between two nodes in can only process the decoded plain data object, but not one hop.

We assign each edge of graph G a positive the encoded data fragment. Then, the path with the In general, each node may have different energy cost shortest transmission time between any two nodes can depending on their energy sources; e. However, the shortest path for any pair of to a constant energy source may have zero energy cost nodes may change over time because of the dynamic while nodes powered by battery may have relatively topology. ETT, considering multiple paths due to nodes high energy cost. Since all tasks Before formulating the problem, we define some func- are instantiated from the same function, we assume tions: 1 f1 i returns 1 if node i in S has at least one they spend approximately the same processing time on task; otherwise, it returns 0; 2 f2 j returns the num- any node.

Given the terms and notations, we are ready ber of instances of task j in S; and 3 f3 z, j returns to formally describe the k-out-of-n data allocation and the transmission cost of task j when it is scheduled for k-out-of-n data processing problems. We note here that T r , the 3.

T r is computed by summing provide any service once it fails. The failure proba- the transmission time in terms of ETT available in D bility of a node estimated at time t is the proba- from node i to its k closest storage nodes of the task. We adopt the the network are selected as processor nodes.

The second remaining energy estimation algorithm in [8] because of constraint Eq 8 indicates that each task is replicated its simplicity and low overhead. Considering that the error each task. The third constraint Eq 9 states that each for estimating the battery remaining time follows a task is replicated at most once to each processor node. Con- nodes. The probability of temporary our framework. Upon receiving the 3.

### Bookseller Completion Rate

Consequently, the delegated In a military application for example, some nodes are node obtains global connectivity information and fail- equipped with better defense capabilities and some ure probabilities of all nodes. This topology information nodes may be placed in high-risk areas, rendering differ- can later be queried by any node. Thus, we define c IEEE.

This type of failure is, however, usually explic- itly known prior to the deployment.

Longer delay im- plies higher transmission energy. As a result, when Fig. The number above each node indicates the minimal transmission time. When we say path p is the failure probability of the node. In our scenario, approximation has to be. The function to be method in a network of 16 nodes. Then, a set of sample graphs can be defined as a multivariate 3. Dij indicating the expected distance between node i Having defined our sample, we determine the number and node j. Another possible case is when all nodes survive usually unknown, so we use the ETT matrix estimator, and either path may be taken.

- Truth, Lies and Deception (The Boy Band Series);
- Yuekun Chen - IEEE Xplore Author Details?
- Energy-Efficient Fault-Tolerant Systems?
- ADHD & ME short story.
- Join Kobo & start eReading today!
- Special order items.
- Table of Contents.

This probability is 0. The expected value estimator and between node 2 and node 3 is In both cases, node 3, 4, 6, 8, 9 are selected as processor nodes and each task is replicated to 3 different processor nodes. The problem is solved in three The k-out-of-n data processing problem is solved in two steps. In the distinct tasks in Rij. Second, we find a schedule with the Task Allocation stage, n nodes are selected as processor minimal energy that has the shortest completion time. An example is shown in Figure 3 a. These two steps are repeated n- instances will be canceled.

In the each iteration. A task can be moved to an earlier time slot relationship between processor nodes and tasks; each as long as no duplicate task is running at the same element Rij is a binary variable indicating whether task time, e. X is a binary vector safely moved to time slot 2 because there is no task containing processor nodes, i.

The objective function minimizes The ILP problem shown in Equations 17 - 20 finds the transmission time for n processor nodes to retrieve M unique tasks from Rij that have the minimal trans- all their tasks. The first constraint Eq 13 indicates mission cost. The second constraint Eq 14 replicates each to be executed on processor node i.

## Mohammad Salehi - Google Scholar Citations

The third Eq 18 ensures that each task is scheduled exactly one constraint Eq 15 ensures that the j th column of R time. The second constraint Eq 19 indicates that Wij can have a non-zero element if only if Xj is 1; and can be set only if task j is allocated to node i in Rij. To solve this min-max problem, nodes, of size less than k, complete all tasks.

The objective function minimizes integer variable Y , 3. Wij The Topology Monitoring component monitors the net- is a decision variable similar to Wij defined previously. Whenever a client node needs cannot consume more energy that the Emin calculated to create a file, the Topology Monitoring component previously. The second constraint Eq 23 schedules provides the client with the most recent topology in- each task exactly once. The third constraint Eq 25 formation immediately. When there is a significant forces Y to be the largest number of tasks on one node. We first give several notations.

A for decision matrix W. Once tasks are scheduled, we term s refers to a state of a node, which can be either U then rearrange tasks — tasks are moved to earlier time and N U. The state becomes U when a node finds that slots as long as there is free time slot and no same task its neighbor table has drastically changed; otherwise, a is executed on other node simultaneously. Algorithm 1 node keeps the state as N U.

We let p be the number depicts the procedure. Note that k-out-of-n data pro- of entries in the neighbor table that has changed. An overview of improved MDFS. Energy measurement setting. The Topology Monitoring component is simple yet 0. The protocol is depicted in Algo- Current Amp 0.

We predefine one node as a topology delegate 0. Upon receiving a beacon message, Fig. Current consumption on Smartphone in different nodes check the IDs in it. For each ID, nodes add the states. When a node needs to access a the ID.

## 9781461441922 - Energy-Efficient Fault-Tolerant Systems (Embedded Systems) by mathews

Our k-out-of-n data allocation allocates file global topology change, the node notifies Vdel ; and Vdel and key fragments optimally when compared with the executes the Topology Discovery protocol. To reduce state-of-art [10] that distributes fragments uniformly to the amount of traffic, client nodes request the global the network. Consequently, our MDFS achieves higher topology from Vdel , instead of running the topology dis- reliability since our framework considers the possible covery by themselves.

Specifically, to evaluate wpa supplicant. We also The experiment was conducted by 8 students who test our k-out-of-n data processing by implementing a carry smartphones and move randomly in an open face recognition application that uses our MDFS. Each file is and the longest node to node distance was 3 hops.