In this talk, I will first give an overview of the research of my lab – the Design of Reliable and Efficient
Algorithms, Models and Systems (DREAMS) Lab – at The Chinese University of Hong Kong. Then I will focus
on one major research theme of characterizing and capitalizing the benefits of looking-ahead in designing
various engineering systems. The investigations are motivated by recent advance of applying machine learning
techniques to leverage historical data to forecast near-future system inputs, e.g., workload for data centers or
wind-power generation in microgrids. Then, as a concrete example, I will elaborate on our recent work on
developing future-aware competitive energy generation strategies in microgrids.
Microgrids represent an emerging paradigm of future electric power systems that integrate both distributed and
centralized generation. Two recent trends in microgrids are the integration of local renewable energy sources
(such as wind farms) and the use of co-generation (i.e., to supply both electricity and heat). However, these
trends also bring unprecedented challenges to the design of intelligent control strategies for the microgrids.
Traditional generation scheduling paradigms assuming perfect prediction of future electricity supply and demand
are no longer applicable to microgrids with unpredictable renewable energy supply and co-generation (that
depends on both electricity and heat demand). In this work, we study online algorithms for the microgrid
generation scheduling problem with intermittent renewable energy sources and co-generation, in order to
maximize the cost savings without the need to predict future demand and supply. Based on insights from the
structure of the offline optimal solution, we propose a class of competitive online algorithms, called CHASE
(Competitive Heuristic Algorithm for Scheduling Energy-generation), that track the offline optimal in an online
fashion. Under typical settings, we show that CHASE achieves the best competitive ratio of all deterministic
online algorithms and the ratio is no larger than 3. We also extend our algorithms to intelligently leverage on
limited prediction of the future, such as near-term demand or wind forecast. By extensive empirical evaluation
using real-world traces, we show that our proposed algorithms can achieve near-offline-optimal performance. In
a representative scenario, CHASE leads to around 20% cost savings with no future look-ahead at all, and the
cost-savings further increase with limited future look-ahead.