Probability Distributions for Channel Utilisation

Abstract

Sensor nets have many undisputed fields of application. A paradigm of communication is the use of one control channel in the MAC layer. We challenge this paradigm for nodes with very restricted hardware resources. In our model nodes support the use of different channels and use clock synchronisation. We present a simple probabilistic synchronised channel utilisation scheme for wireless communication. The main features are its simplicity, robustness against radio interference, the high throughput caused by less interfering signals, and predictable energy consumption. For this, the channel selection is based on a carefully chosen probability distribution maximising the expected number of successfully delivered packets up to a constant factor. Combined with a standard synchronisation scheme it provides a novel energyefficient, robust, and fast message delivery service for sensor networks where data gathering is not available due to memory restrictions.

Oblivious parallel probabilistic channel utilization without control channels

“…For this we have proved an upper bound on the transmitted messages for the single round channel utilization protocol if the number of senders and receivers is unknown. This upper bound solves the recently stated open problem by the authors of this paper in [SV05]. The conclusion is that the proposed factorized geometric probability distribution for channel utilization in the single round model in [SV05] is in fact asymptotically optimal.…”

“…Combining this upper bound with the result of [SV05] completes the proof at the end of this subsection.…”

“…Section 4 presents and analyzes the Funnel protocol, the main result of this paper. Also the open question of [SV05] is solved here: whether the s-factorized geometric probability distribution is optimal for the single round model. In the last section a conclusion is given.…”

“…It is a paradigm in sensor networks to use one channel for communication. In [SV05] we have already presented a new concept in which one of several frequencies is probabilistic selected. There, the expected number of delivered messages in one sending attempt was optimized.…”

“…The task of channel selection and utilization of the communication medium is the classical problem of the Medium Access Layer which is positioned in the Data Link Layer in the OSI Model (see IEEE P802.3ae). Using an uni-form probability distribution for selecting the communication channel (for sending or receiving actions) performs best if the number of senders is equivalently to the number of channels [SV05]. Unfortunately, the number of senders and receivers for each sending attempt depends on many factors (number of nodes in sending range, probability for a message, probability of failure in previous attempts .…”