Session details: Selected extended papers from SBESC 2012 Brazilian symposium on computing system engineering

Abstract

The Brazilian Symposium on Computing System Engineering (SBESC) is an initiative of the research community originally associated with three events: the Brazilian Workshop on Real-Time Systems, created in 1998; the Brazilian Workshop on Operating Systems, created in 2004; and the Brazilian Workshop on Embedded Systems, created in 2010. The identification of a strong synergy among these research areas added to the fact that designing computing systems is an increasingly multidisciplinary task has motivated the workshops to move from their native conferences to form an independent symposium. The broad term "Computing System Engineering" emerged in the context of contemporary Cyber-Physical Systems, which challenge Engineers and Scientist as they shape a new reality for modern societies. This year, the symposium was hosted by the Federal University of Rio Grande do Norte (UFRN) in the beautiful coastal city of Natal in Northeast Brazil. The program spanned a whole week, from November 5 to 9, and was organized in two big blocks: peer-reviewed paper sessions and tutorials. The event was organized by UFRN's Computer Science Department (DIMAP), sponsored by the Brazilian Computing Society (SBC), and financed by CAPES, CNPq, and FAPERN. SBESC's Technical Program Committee also selected two papers from those presented at the symposium addressing operating systems issues to appear in the Operating Systems Review and thus offer ACM fellows a glimpse of the event. Authors were invited to extend and improve their manuscripts with recent research results. The first paper, On the Formal Verification of Component-based Embedded Operating Systems, by M. Ludwich and A. Frohlich, introduces an approach to formally verify the functional correctness and safety properties of embedded operating system components described at system-level. The convergence between software and hardware in the domain of embedded systems is pressing operating systems in the filed to deliver their services both as software and as hardware. In such a scenario, it is desirable to verify system properties regardless of whether components are instantiated as software or hardware. The approach combines software model checking and programming by contract concepts to address such issues. The second, Adaptive Request Batching for Byzantine Replication, by A. de Sa, A. Freitas, and R. Macedo, proposes an adaptation of the Practical Byzantine Fault Tolerance algorithm proposed by Liskov and Castro to increase its performance. The original PBFT algorithm uses a batch to store client requests on the primary replica. Requests in a batch are handled together, aiming at decreasing the time spent with message authentication. The adaptation strategy dynamically varies the batch size and the batch time-out according with application activity.