ABSTRACT
Self-healing gradients are distributed estimates of the distance from each device in a network to the nearest device designated as a source, and are used in many pervasive computing systems. With previous self-healing gradient algorithms, even the smallest changes in the source or network can produce small estimate changes throughout the network, leading to high communication and energy costs. We observe, however, that in many applications, such as routing and geometric restriction of processes, devices far from the source need only coarse estimates, and that a device need not communicate when its estimate does not change. We have therefore developed Flex-Gradient, a new self-healing gradient algorithm with a tunable trade-off between precision and communication cost. When distance is estimated using Flex-Gradient, the constraints between neighboring devices are flexible, allowing estimates to vary by an amount proportional to a device's distance to the source. Frequent small changes in the network or source thus cause frequent estimate changes only within a distance proportional to the magnitude of the change, as verified in simulation on a network of 1000 devices. This can enable drastic reductions in the communication and energy cost of gradient-based algorithms.
- J. Bachrach and J. Beal. Programming a sensor network as an amorphous medium. In Distributed Computing in Sensor Systems (DCOSS) 2006 Poster, June 2006.Google Scholar
- J. Bachrach, R. Nagpal, M. Salib, and H. Shrobe. Experimental results and theoretical analysis of a self-organizing global coordinate system for ad hoc sensor networks. Telecommunications Systems Journal, Special Issue on Wireless System Networks, 2003.Google Scholar
- J. Beal, J. Bachrach, D. Vickery, and M. Tobenkin. Fast self-healing gradients. In ACM Symposium on Applied Computing, March 2008. Google ScholarDigital Library
- W. Butera. Programming a Paintable Computer. PhD thesis, MIT, 2002. Google ScholarDigital Library
- L. Clement and R. Nagpal. Self-assembly and self-repairing topologies. In Workshop on Adaptability in Multi-Agent Systems, RoboCup Australian Open, Jan. 2003.Google Scholar
- Q. Fang, J. Gao, L. Guibas, V. de Silva, and L. Zhang. Glider: Gradient landmark-based distributed routing for sensor networks. In INFOCOM 2005, March 2005.Google ScholarCross Ref
- C. Intanagonwiwat, R. Govindan, and D. Estrin. Directed diffusion: A scalable and robust communication paradigm for sensor networks. In Sixth Annual International Conference on Mobile Computing and Networking (MobiCOM '00), August 2000. Google ScholarDigital Library
- H. Luo, F. Ye, J. Cheng, S. Lu, and L. Zhang. Ttdd: A two-tier data dissemination model for large-scale wireless sensor networks. Journal of Mobile Networks and Applications (MONET), 2003.Google Scholar
- M. Mamei, F. Zambonelli, and L. Leonardi. Co-fields: an adaptive approach for motion coordination. Technical Report 5--2002, University of Modena and Reggio Emilia, 2002.Google Scholar
- F. Ye, G. Zhong, S. Lu, and L. Zhang. Gradient broadcast: a robust data delivery protocol for large scale sensor networks. ACM Wireless Networks (WINET), 11(3): 285--298, 2005. Google ScholarDigital Library
Index Terms
- Flexible self-healing gradients
Recommendations
Fast self-healing gradients
SAC '08: Proceedings of the 2008 ACM symposium on Applied computingWe present CRF-Gradient, a self-healing gradient algorithm that provably reconfigures in O(diameter) time. Self-healing gradients are a frequently used building block for distributed self-healing systems, but previous algorithms either have a healing ...
Fast Self-stabilization for Gradients
DCOSS '09: Proceedings of the 5th IEEE International Conference on Distributed Computing in Sensor SystemsGradients are distributed distance estimates used as a building block in many sensor network applications. In large or long-lived deployments, it is important for the estimate to self-stabilize in response to changes in the network or ongoing ...
Stochastic gradient MCMC with stale gradients
NIPS'16: Proceedings of the 30th International Conference on Neural Information Processing SystemsStochastic gradient MCMC (SG-MCMC) has played an important role in large-scale Bayesian learning, with well-developed theoretical convergence properties. In such applications of SG-MCMC, it is becoming increasingly popular to employ distributed systems, ...
Comments