Call for Papers
Motivation
- timing of the input, output, or both;
- minimal quality of the delivered output/service;
- limited energy, power, or processing/communication resources, and more.
- the size of the problem (e.g. the sample automotive application of the WATERS 2017 challenge composed by ~1000 functions and ~10000 shared variables);
- the accuracy of the model for the workload and the computing platform;
- the tolerable degree of approximation of the constraints.
Topics of Interest
- Optimization variables:
- parameters of tasks (period, execution times, etc.);
- amount of processing and communication capacity (speed/bandwidth);
- heterogeneous computing: type/quantity of accelerators (GPU, FPGA elements, etc).
- Model of the cost of design:
- borrowed from different application contexts: automotive, avionics, data centers, mobile devices, edge;
- maximum extensibility of functionalities;
- maximum battery savings and lifetime.
- Model of the constraints:
- schedulability constraints;
- sensitivity analysis;
- deadline model: hard/soft/probabilistic constraints;
- accurate vs. efficient representation of constraints: approximations.
- Solvers:
- continuous vs. discrete methods (granularity of discretization);
- gradient-descent, linear programming (LP), quadratic programming;
- mixed-integer linear programming (MILP);
- constraint programming (CP);
- meta-heuristics: simulated annealing, tabu search, genetic algorithms;
- efficient problem formulations.
Type of submissions
- Original contributions.
- Preliminary ideas, not necessarily mature, seeking for feedback from the community.
- Short versions of previously published research, relevant to the field.