@article {baquero1998improving, title = {Improving causality logging in mobile computing networks}, journal = {Mobile Computing and Communications Review}, volume = {2}, number = {4}, year = {1998}, pages = {62{\textendash}66}, publisher = {ACM}, abstract = {

This paper builds on previous techniques for efficient causality logging in mobile networks and presents a lighter logging mechanism. The technique is based on a particular partial order that is generated by the interleaving of events on mobile hosts that are mediated by the same support station. I. Introduction Mobile computing systems are frequently designed as a network of fixed nodes, mobile support stations (MSS), that give connectivity to a set of mobile hosts (MH) [1]. The MSSs are interconnected by a high bandwidth wire-line network where communication costs are inexpensive. In contrast, the MHs always communicate with the mediation of a hosting MSS using a low-bandwidth wireless or phone channel where costs are at issue. This class of mobile computing systems, although excluding direct communication among MHs, models a vast range of existing systems that include wireless networks of MHs bound to local cells, and nomadic MHs that bind to different MSSs as access points to a wide area network. Distributed applications that build on this class of mobile computing systems are often modeled as a set of concurrent activities distributed among different MHs. Tracking the causal relationships among these concurrent activities is a basic mechanism for the analysis and debugging of distributed applications and a step towards the design of message delivery and replica consistency policies. It is well established [7] that in a distributed system the causal dependency can be fully characterized by the use of vector clocks [7, 2, 5]. However vector clocks are very sensitive to scalability issues since the vector size is bound to the number of activities, which are here bound to the number of MHs. Clearly, a dependency tracking mechanism that is bound to the number of MSSs an.

}, attachments = {https://haslab.uminho.pt/sites/default/files/cbm/files/10.1.1.41.8136.pdf}, author = {Carlos Baquero Moreno and Francisco Moura} }