Thomas Wolcott

Intellectual Merit. The ability of microscopic larvae to control their fate has long been debated. Larvae often are assumed to be at the mercy of currents, typically being carried far from natal populations and recruiting unpredictably in time and space. It has become increasingly evident that larvae of coastal species strongly influence their transport; some remain within estuaries, while others make true migrations between adult and larval habitats, even out to the edge of the continental shelf. Rates and directions of larval transport are thought to be determined largely by the timing, duration and amplitude of vertical migrations, and mean depth larvae occupy, in stratified flow. We know little about how different behaviors affect dispersal trajectories, alongshore transport, distribution of recruits in time and space, and the connectivity of populations; new approaches are needed. Under NSF and EPA funding Wolcott & Wolcott have developed and field tested a drifter that responds to its physical environment by mimicking larval vertical migration behaviors . We now propose to provide one of the first direct tests of how effectively behavior enables larvae to complete cross-shelf migrations and limit their alongshore dispersal. The study will be conducted in a region of persistent upwelling where strong currents are widely believed to overwhelm larval swimming and limit recruitment to adult populations. This challenge is intensified at upwelling centers and jets where currents are strongest, leading to the view that upwelling shadows in the lee of capes are the main refuge for developing larvae. Contrary to this prevailing paradigm, under prior NSF funding we (Morgan) showed that larvae of many nearshore species remain very close to shore (<3 km) at high densities, while others migrate across the shelf even during strong upwelling conditions. We then entered documented interspecific differences in larval behavior into a ROMS model, which predicted cross-shelf distributions similar to those observed in the field and gave the first realistic estimate of alongshore dispersal and population connectivity. However, ROMS models do not resolve the innermost shelf where most larvae develop, hence probably underestimate the extent of local retention. To address this, we will track biomimetic (ABLE) drifters on the innermost shelf many times during upwelling-relaxation cycles over 3 yr to contrast the effect of four simulated behavior patterns on hypothesized behavioral-physical mechanisms of larval transport, supply and connectivity. We will test three fundamental hypotheses that all bear on maintenance of biocomplexity. 1. Specific larval behaviors lead to retention of larvae near shore. 2. Specific larval behaviors lead to decreased longshore dispersal of larvae. 3. Different larval behaviors have pronounced effects on retention and transport at oceanographic features like upwelling centers, jets and shadows. The underlying question: in light of our findings, how should dimensions and placement of Marine Protected Areas (MPAs) be optimized? Broader Impacts. The proposed research will provide the first experimental field test of how larval behavior affects the rates, directions and distances of transport and population connectivity in upwelling regimes. We will compare direct measurements from ABLE drifters with both observed and modeled cross-shelf larval distributions, and with modeled alongshore transport. We expect that 1) larval behaviors will have starkly different consequences on retention and export from upwelling cells, 2) larvae commonly remain much closer to home than has been thought, and 3) effective MPAs could be smaller and placed more closely together. Knowledge of underlying mechanisms regulating larval transport , and hence the location and magnitude of recruitment, is scant but is central to our understanding of the ecology and evolution in the sea and our ability to anticipate the impacts of climate change on marine populations and communities. Our results will be broadly applicable to

The proposed work will improve memory systems for high-performance real-time embedded systems. We will develop methods implemented in hardware and software to create fast, efficient memory systems with easily predicted performance. This is critical to designers of systems with real-time requirements, as memory access time affects task execution time. A task's worst-case execution time (WCET) is the fundamental metric which is used in equations to determine how to make task scheduling decisions. For most non-trivial hardware and software, computing the exact WCET is impossible. This leads designers to estimate a safe WCET bound which is guaranteed by design to never be smaller than the actual WCET. The closer the WCET bound is to the WCET, the more efficient the system can be, as there is less time dedicated to a margin of safety. Unfortunately, it is difficult to analyze the timing impact of caches and virtual memory systems on a task's WCET bound, much less WCET, resulting in overly conservative WCET bound estimates which may make the use of cache and virtual memory impractical. The goal of this proposal is to investigate methods to improve cache and virtual memory performance in a way that is easy to predict.

We propose to develop new methods and a framework to allow users to quickly implement efficient software-based controllers for customized network communication protocols. The wide variety of embedded systems communication requirements leads to numerous different embedded networking protocols -- a single generic solution would lead to excessive inefficiency in performance, cost, power, reliability, code size, and other areas. Automobiles, trucks, trains, elevators, aircraft, ships, security systems, factories, office building, sensor networks, and military vehicles all have varying communications requirements, leading to a variety of communication protocols. Networking protocol implementations using only traditional software methods for scheduling and context switching are often inadequate due to high timing overhead and variability. A dedicated hardware implementation requires the design of an integrated circuit, discrete logic or the use of programmable logic such as FPGA, and it often lacks the flexibility of a software solution. We proposed to develop methods, a toolbox and an associated communication framework to allow users to quickly implement software-based controllers for customized network communication protocols. More specifically, we will provide a complete networking stack featuring several options at each layer in the stack. Users will select specific protocol characteristics, and the tools of the framework will generate (and compile) the code that implements the specific protocol options for the desired application. C17A major component of the proposed work is evaluating the performance of the customized network stack. In the first phase, corresponding to different choices of the end user, the tool will provide an estimation of the performance of the proposed protocol stack: efficiency, delay, throughput, power consumption, expected error rates, etc. Since many performance parameters depend on the particular deployment scenario, we will also design an emulator that will enable an exact evaluation of the performance of the embedded system in the targeted environment. The goal of the performance evaluation component is to allow the designer to make the correct decisions will be made at the early stages in the product development. C2The proposed tool does not specifically target any class of embedded systems; however, we recognize that a solution with less resources will typically be preferrable not only due to monetary cost, but also size, energy and thermal constraints and hence we will seek efficient solutions whenever possible. Therefore, we will use software thread integration to provide efficient concurrency. Our work will allow practitioners to implement optimized protocols, by providing the design and analysis tools that bridge the gap between network and CPU simulation. More importantly, we foresee that a large percentage of the users of our system will be non-specialists. We will specifically provide custom-fit protocols that are needed in many non-communication engineering fields, such that a communication system will be much easier to design and debug. While the proposed tool will be very general and target a broad range of hardware (microcontrollers and transceiver), we will use two sample applications from two such unrelated field as a reality check mechanism. Specifically, we will consider ultrasonic biotelemetry, a powerful new research tool for analyses of movement, habitat utilization, physiological function and behavior of marine organisms as one of such applications of networking in unrelated fields. Furthermore, a structural health monitoring of reinforced bridges using wireless sensor networks will be a second application that we will consider when designing the general purpose networking protocol toolbox. We believe that many non-specialists in unrelated fields, but in need of communication protocols will benefit directly from the results of this project. Even the researchers in