Abstract:
Low energy consumption, power dissipation and fault tolerance area unit typically key objectives within the style of real-time embedded systems. Real-time systems sometimes use system level energy reduction ways Postponement technique supported DPM is employed within which the processors area unit run at the most offer voltage, and might tolerate one transient fault. It will operate with 2 cases with low employment and high employment. Within the case of low employment it's enough to shelve the secondary copy of method to a time when finishing the first process, if the first copy finishes with success, doesn't need corporal punishment any portion of the secondary method. Within the case of high employment, the secondary copy should begin corporal punishment before the first copy finishes, as a result of otherwise if a fault happens throughout the execution of the first copy, there'll be no sufficient time left to execute the secondary copy before the point in time. To fulfill the point in time, the secondary copy are often deferred at the most (D-ζ) units of your time once no fault happens, the first copy is dead fully and also the secondary copy is dead for (2ζ-D) units of your time. There's no got to execute the remaining (D-ζ) of the secondary copy, unless the first copy becomes faulty. During this work, we tend to propose Dynamic Cache Reconfiguration (DCR) exploitation runtime observation of cache parameters to additional improve the energy potency. On merging Postponement technique supported DPM and DCR with leak aware formula is employed which may minimize the general system energy consumption and leak power by considering the temperature. Simulation results show reduction of energy dissipation up to 16% compared to different techniques. The results additionally indicate that for lower set associativity and better task generation, the projected approaches have higher performance
