Optimal energy efficient packet scheduling with arbitrary individual deadline guarantee

Given a rate adaptive wireless transmitter, a challenging problem is to design a rate control policy for it such that the energy consumption is minimized at transmitting a set of dynamically arrived packets with arbitrary individual deadlines. In a decade, researches have partially made progress on this topic. A latest work offers an optimal algorithm that allows packets to have arbitrary deadlines but requires them to follow the order they arrive. This paper first presents the Densest Interval First (DIF) policy which repeatedly locates the densest data interval and determines its transmission rate. This policy is proved to be optimal for the most general model that allows arbitrary arrival times as well as arbitrary deadlines. Then, this paper presents a simple EDF (earliest deadline first) algorithm to actually schedule the transmission time for each packet. It is proved that the EDF always guarantees every packet to complete transmission before its deadline with minimum energy consumption which is computed and required by DIF. Finally, this paper also proposes a novel online policy named Density Guided Cooling (DGC) policy which models Newton's Law of Cooling. Simulations show that online DGC policy constantly produces a rate scheduling that on average consumes energy within 110% of the minimum value obtained by the offline DIF.