An Approach for Detecting Power Peaks During Testing and Breaking Systematic Pathological Behavior

Abstract

The verification and validation process of embedded critical systems requires providing evidence of their functional correctness and also that their non-functional behavior stays within limits. In this work, we focus on power peaks, which may cause voltage droops and thus, challenge performance to preserve correct operation upon droops. In this line, the use of complex software and hardware in critical embedded systems jeopardizes the confidence that can be placed on the tests carried out during the campaigns performed at analysis. This is so because it is unknown whether tests have triggered the highest power peaks that can occur during operation and whether any such peak can occur systematically. In this paper we propose the use of randomization, already used for timing analysis of real-time systems, as an enabler to guarantee that (1) tests expose those peaks that can arise during operation and (2) peaks cannot occur systematically inadvertently.

Worst Case Execution Time and Power Estimation of Multicore and GPU Software: A Pedestrian Detection Use Case

“…Again the software randomisation of the unit under analysis is used in order to generate power peaks that might not be observed during the measurements performed at analysis, e.g. due to the increased number of misses generated by a given memory layout, and predict its probabilistic worst case using Extreme Value Theory, similar to [7]. In addition, the configurable contenders allow to upperbound the power consumption of the activity in the other cores.…”

“…Such an estimation is useful in order to guarantee the adequate thermal design of the system, so that it stays within its operational limits, otherwise there is a risk of unavailability of the safety critical system. However, the complexity of the hardware complicates this estimation using measurement based techniques [7], since during the experiments performance during analysis time, there are power measurement peaks which might not be observed.…”

“…Usually manufacturers provide a conservative upperbound of the worst case power consumption, known as Thermal Design Power (TDP), however this can result in overprovisioning of the system resources. The use of Extreme Value Theory (EVT) [8], which is one of the foundations of MBPTA, has been shown possible in the past for the worst case power consumption of a simulated time-randomised multicore CPU based on the NGMP [7]. Similar to MBPTA, hardware or software randomisation is required to enable the use of this method.…”