It is increasingly important to structure signal processing algorithms and systems to allow for trading off between the accuracy of results and the utilization of resources in their implementation. In any particular context, there are typically a variety of heuristic approaches to managing these tradeoffs. One of the objectives of this paper is to suggest that there is the potential for developing a more formal approach, including utilizing current research in Computer Science on Approximate Processing and one of its central concepts, Incremental Refinement. Toward this end, we first summarize a number of ideas and approaches to approximate processing as currently being formulated in the computer science community. We then present four examples of signal processing algorithms/systems that are structured with these goals in mind. These examples may be viewed as partial inroads toward the ultimate objective of developing, within the context of signal processing design and implementation,...

superclass for all monitor control policy objects. PRIORITYCEILINGEMULATION 87 6.1.1 Constructors public Monitor ontrt () 6.1.2 Methods public static void setMonitor Contr l(MonitorControl8 policy) Control the default monitor behavior for object monitors used by synchronized statements and methods in the system. The type of the policy object determines the type of behavior. Conforming implementations must support priority ceiling emulation and priority inheritance for fixed priority preemptive threads. Parameters: policy - The new monitor control policy. If null nothing happens. public static void setMonitor Contr l(java.lang.Object monitor MonitorControl 8 policy) Has the same effect as setMonitorControl(), except that the policy only affects the indicated object monitor. Parameters: monitor - The monitor for which the new policy will be in use. The policy will take effect on the first attempt to lock the monitor after the completion of this method. If null nothing wi...

n a hard real-time cyctem, every time-T critical task must meet its timing con-1 straint, typically specified as its deadline. (A task is a granule of computa-tion treated by the scheduler as a unit of work to be allocated processor time, or scheduled.) If any time-critical task fails to complete and produce its result by its deadline, a timing fault occurs and the task's result is of little or no use. Such factors as variations in the processing times of dynamic algorithms make meeting all deadlines at all times difficult. The imprecise computation te~hniquel-~ can minimize this difficulty. It prevents timing faults and achieves graceful degra-dation by giving the user an approximate result of acceptable quality whenever the system cannot produce the exact result in time. Image processing and tracking are examples of real-time applications where the user may accept timely approximate results: Frames of fuzzy images and rough estimates of location produced in time may be better than perfect images and accurate location data produced too late. In this article, we review workload mod-els that quantify the trade-off between re-sult quality andcomputation time. We also describe scheduling algorithms that ex-ploit this trade-off.

Scheduling theory holds great promise as a means to a priori validate timing correctness of real-time applications. However, there currently exists a wide gap between scheduling theory and its implementation in operating system kernels running on specific hardware platforms. The implementation of any particular scheduling algorithm introduces overheadand blocking components which must be accounted for in the timing correctness validation process. This paper presents a methodology for incorporating the costs of scheduler implementation within the context of fixed priority scheduling algorithms. Both event-driven and timerdriven scheduling implementations are analyzed. We show that for the timer-driven scheduling implementations the selection of the timer interrupt rate can dramatically affect the schedulability of a task set, and we present a method for determining the optimal timer rate. We analyzed both randomly generated and two well defined task sets and found that their schedulabil...

This paper proposes and investigates an approach to deduction in probabilistic logic, using as its medium a language that generalizes the propositional version of Nilsson's probabilistic logic by incorporating conditional probabilities. Unlike many other approaches to deduction in probabilistic logic, this approach is based on inference rules and therefore can produce proofs to explain how conclusions are drawn. We show how these rules can be incorporated into an anytime deduction procedure that proceeds by computing increasingly narrow probability intervals that contain the tightest entailed probability interval. Since the procedure can be stopped at any time to yield partial information concerning the probability range of any entailed sentence, one can make a tradeoff between precision and computation time. The deduction method presented here contrasts with other methods whose ability to perform logical reasoning is either limited or requires finding all truth assignments consistent ...

This thesis develops the Sporadic Server (SS) algorithm for scheduling aperiodic tasks in real-time systems. The SS algorithm is an extension of the rate monotonic algorithm which was designed to schedule periodic tasks. This thesis demonstrates that the SS algorithm is able to guarantee deadlines for hard-deadline aperiodic tasks and provide good responsiveness for soft-deadline aperiodic tasks while avoiding the schedulability penalty and implementation complexity of previous aperiodic service algorithms. It is also proven that the aperiodic servers created by the SS algorithm can be treated as equivalently-sized periodic tasks when assessing schedulability. This allows all the scheduling theories developed for the rate monotonic algorithm to be used to schedule aperiodic tasks. For scheduling aperiodic and periodic tasks that share data, this thesis defines the interactions and schedulability impact of using the SS algorithm with the priority inheritance protocols. For scheduling ha...

This paper describes an organizing conceptual structure for current real-time AI research, clarifying the different meanings this term has acquired for various researchers

Admission control and overload management techniques are central to the design and implementation of RealTime Database Systems. In this paper, we motivate the need for these mechanisms and present protocols for adding such capabilities to Real-Time Databases. In particular, we present a novel admission control paradigm, we describe a number of admission control strategies and contrast (through simulations) their relative performance. 1 Introduction The main challenge involved in scheduling transactions in a Real-Time DataBase (RTDB) management system is that the resources needed to execute a transaction are not known a priori. For example, the set of objects to be read (written) by a transaction may be dependent on user input (e.g. in a stock market application) or dependent on sensory inputs (e.g. in a process control application). Therefore, the a priori reservation of resources (e.g. read/write locks on data objects) to guarantee a particular Worst Case Execution Time (WCET) bec...

We describe a real-time speech application which simulates the human production of natural-language utterances in a two-person conversation. Since generating such utterances is a complex and often timeconsuming task, we examine the timing constraints and result-quality requirements of this application, as well as the dependencies between the timing constraints and result qualities of the tasks that generate utterances. We then extend the imprecise computation model to capture these dependencies and discuss the general applicability of the extended model. 1 Introduction The imprecise computation technique [3] is a way to improve the scheduling feasibility of applications where results of poorer quality are better than late results. The imprecise computation model assumes that the deadline of each task is given and that the quality of a task's result is solely dependent on the time and resources spent to produce the result. However, many real-time applications cannot be adequately char...

The incorporation of unbounded components (i.e. software modules that cannot be analysed to produce realistic worst case execution times) into hard real-time applications has been recognised as a key issue for the next generation of systems. We present an architectural model that caters for the three main approaches to integrating unbounded components --- imprecise computation, sieve functions and multiple versions. This architectural model is feasible because it is supported by schedulability tests that will guarantee the bounded tasks. These test are defined in the paper. A simple computational model that uses preemptive priority-based dispatching is required. The wide-spread use of techniques such as imprecise computation will only happen if they are integrated into standard software engineering methods. We therefore show how the techniques can be realised in Ada 9X. ############### + This work is supported, in part, by the UK Information Engineering Advanced Technology Programme, ...