Ismael Ripoll 
Category: Assistant Professor (Titular de Universidad)
Phone: +34 963 879 218 / Internal: 79218 / Fax+34 963 877 579
Location: Room 3N12, Building 1G, Universitat Politècnica de València.
  • Hyper-Periods
  • Scheduling
  • Dynamic memory
  • Security
Research is both: fun and useful !

First of all: What is the hyper-period?: The hyper-period is the smallest interval of time after which the periodic patterns of all the tasks are repeated. It is typically defined as the LCM (least common multiple) of the periods of the tasks (in a periodic task system).

A small hyper-period value has several applications in several fields of real-time scheduling:

  • lowering complexity in table driven schedulers,
  • reducing search space in model checking analysis,
  • generating synthetic workload for statistical analysis of real-time scheduling algorithms, etc.

There is little room form improvements when using the mathematical definition of hyper-period, it is like trying to reduce the result of 5 * 5 because you think that 25 is too big

But what would occur if the periods were not an integer number but defined as a range of valid values. That is, a period is defined as a nominal value and a tolerance. In fact, this is the normal way of dealing with the physical magnitudes commonly used by engineers and physics.

What is , is that with this "engineering" definition of period, it is possible to exponentially reduce the value of the resulting hyper-period.


The question to answer is: Given a periodic task set. Is it feasible?

This problem is the first issue that must be addressed if we want to use the EDF in a real-time system. Once we know how to analyse the schedulability of a basic task set, then it is possible to extend the solution to include more restrictions like precedence constrains between tasks, context switch overhead, or mutual exclusive resources.

The first solution to this problem was propossed in 200? by San joy Barhua. Three years latter I proposed another solution based on a completely different property. This new solution paved the way for the optimal aperiodic service for EDF.


The dynamic memory allocation (malloc/free) is a technique widely used since the very beginning of the computer science. It was deeply studied during the 60 and 70, mainly with the intention to address the fragmentation problem. Later the goal was to speed-up the temporal cost of the operations (allocate and de-allocate). Despite the many research efforts, it is still an open problem that is waiting to be better settled.

A mayor breakthrough was achieved a few years ago: the design of the TLSF (Two Level Segregate Fit) allocator. It is a fast, constant time allocator with a very low fragmentation.

Basically, there are two problems regarding dynamic memory: fragmentation and temporal efficiency. There exists many misconceptions around DMA, one of them was that lower fragmentation can be achieved by more complex and costly algorithms. For example it is generally accepted that the policy "best-fit" causes less fragmentation, but it is not true!.


Is it possible to "remove" software bugs using the virtualisation?

In short, yes as far as different processor architectures behalves diferently to buggy code. Therefore, if we compile our application for serveral target processors, it is likely that in some of them the programming errors do not manifest.

Although it may seem a complex task (cross compiler suites, different processors, !?!?), thanks to the enormous efforts done by the open source community in the development of the GNU GCC toolchain, and the incredible work of Fabrice Bellard with the Qemu emulator, it is almost trivial.

Teaching Currently I'm teaching: Phd student:
  • Vicent Fernandez
[1] I. Ripoll and R. Ballester-Ripoll, Period selection for minimal hyperperiod in periodic task systems, Computers, IEEE Transactions on, 62 (2013), pp. 1813-1822. [ bib | DOI ]
[2] Emerging trends in ICT security, Elsevier Inc, Dic 2013, ch. 21, pp. 335-357. [ bib | DOI ]
[3] P. Balbastre, I. Ripoll, and A. Crespo, Period sensitivity analysis and d-p domain feasibility region in dynamic priority systems, Journal of Systems and Software, 82 (2009), pp. 1098-1111. [ bib | DOI ]
[4] M. Masmano, I. Ripoll, P. Balbastre, and A. Crespo, A constant-time dynamic storage allocator for real-time systems, Real-Time Systems, 40 (2008), pp. 149-179. [ bib | DOI ]
[5] M. Masmano, I. Ripoll, J. Real, A. Crespo, and A. J. Wellings, Implementation of a constant-time dynamic storage allocator, Softw., Pract. Exper., 38 (2008), pp. 995-1026. [ bib | DOI ]
[6] P. Balbastre, I. Ripoll, and A. Crespo, Minimum deadline calculation for periodic real-time tasks in dynamic priority systems, IEEE Trans. Computers, 57 (2008), pp. 96-109. [ bib | DOI ]
[7] A. Crespo, I. Ripoll, M. G. Harbour, and G. Lipari, Operating system support for embedded real-time applications, EURASIP Journal on Embedded Systems, 2008 (2008). [ bib | DOI ]
[8] P. Balbastre, I. Ripoll, J. Vidal, and A. Crespo, A task model to reduce control delays, Real-Time Syst., 27 (2004), pp. 215-236. [ bib | DOI ]
[9] I. Ripoll, A. Crespo, and A. García-Fornes, An optimal algorithm for scheduling soft aperiodic tasks in dynamic-priority preemptive systems, IEEE Transactions on Software Engineering, 23 (1997), pp. 388-400. [ bib | DOI ]
[10] I. Ripoll, A. Crespo, and A. K. Mok, Improvement in feasibility testing for real-time tasks, Real-Time Syst., 11 (1996), pp. 19-39. [ bib | DOI ]
[1] H. Marco-Gisbert and I. Ripoll, Preventing brute force attacks against stack canary protection on networking servers, in Network Computing and Applications (NCA), 2013 12th IEEE International Symposium on, 2013, pp. 243-250. [ bib | DOI ]
[2] V. Brocal, P. Balbastre, R. Ballester, and I. Ripoll, Task period selection to minimize hyperperiod, in Proceedings of the 16th IEEE International Conference on Emerging Technologies and Factor Automation, September 2011. [ bib ]
[3] Á. Esquinas, J. Zamorano, J. A. de la Puente, M. Masmano, I. Ripoll, and A. Crespo, Ork+/xtratum: An open partitioning platform for ada, in Ada-Europe, 2011, pp. 160-173. [ bib | DOI ]
[4] M. Masmano, I. Ripoll, S. Peiró, and A. Crespo, Xtratum for leon3: an open source hypervisor for high integrity systems, in European Conference on Embedded Real Time Software and Systems. ERTS2 2010., Toulouse (France), May 2010. [ bib ]
[5] A. Crespo, I. Ripoll, and M. Masmano, Partitioned embedded architecture based on hypervisor: The xtratum approach, in EDCC, 2010, pp. 67-72. [ bib | DOI ]
[6] C. Jouvray, G. Chartier, N. François, I. Ripoll, M. Masmano, and A. Crespo, Enforcing trust in control automotive platforms, in Proceedings of the 1st Workshop on Critical Automotive applications: Robustness & Safety, CARS '10, New York, NY, USA, 2010, ACM, pp. 43-46. [ bib | DOI ]
[7] M. Masmano, I. Ripoll, A. Crespo, and J. Metge, Xtratum: a hypervisor for safety critical embedded systems, in Eleventh Real-Time Linux Workshop, Dresden (Germany), September 2009. [ bib ]
[8] M. Masmano, I. Ripoll, A. Crespo, J. Metge, and P. Arberet, Xtratum: An open source hypervisor for TSP embedded systems in aerospace, in DASIA 2009. DAta Systems In Aerospace., May. Istanbul 2009. [ bib ]
[9] P. Balbastre, I. Ripoll, and A. Crespo, Exact response time analysis of hierarchical fixed-priority scheduling, in RTCSA, 2009, pp. 315-320. [ bib | DOI ]
[10] A. Marchand, P. Balbastre, I. Ripoll, and A. Crespo, Providing memory qos guarantees for real-time applications, in RTCSA, 2008, pp. 312-317. [ bib | DOI ]
[11] S. Peiro, M. Masmano, I. Ripoll, and A. Crespo, PaRTiKle OS, a replacement of the core of RTLinux, in 9th Real-Time Linux Workshop, 2007. [ bib ]
[12] M. Masmano, I. Ripoll, and A. Crespo, An overview of the xtratum nanokernel, in Workshop on Operating Systems Platforms for Embedded Real-Time applications, 2005. [ bib ]
[13] M. Masmano, I. Ripoll, A. Crespo, and J. Real, Tlsf: A new dynamic memory allocator for real-time systems, in ECRTS, 2004, pp. 79-86. [ bib | DOI ]