Fragmentation of data packets in wireless sensor network with variable temperature and channel conditions

Abstract

This study formalizes and solves the problem of minimizing the energy dissipated by a node in order to successfully transmit the initial data amount during one round of communication over a wireless channel affected by Rice fading under varying temperature conditions. In this work, the models for the temperature dependence of the parameters in an incoherent frequency shift key (FSK) demodulator are presented and investigated for the first time, taking into account both the frequency separation and the synchronization type of the clock generator. The studied orthogonality coefficients of FSK signals represent internal parameters in the proposed model of the incoherent Rice wireless channel. This model allows the analytical relationship between the bit error probability (BEP) and node heating temperature, signal fading depth in the channel (Rician K-factor), and signal-to-noise ratio (SNR) to be established. Since it is necessary to retransmit the entire data packet when an erroneous bit is detected in a communication system that uses an automatic repeat request, at high BEP values, it is recommended to fragment packets in order to improve the energy efficiency of the nodes. By using the developed models for calculating energy losses during the communication round, it was determined that optimal fragmentation options (OFO) for data packets can be identified within specific ranges of temperature and Rician K-factor, which includes the overhead costs for packeting and retransmission. As a result, OFO matrices of packets in the “temperature – K-factor” (t0–K) coordinate system were obtained for minimizing the energy loss of nodes at different SNR values. In our example of using TI CC2500 node transceivers for 90-day monitoring of corn vegetation, the t0K-Frag algorithm reduced node energy losses by a factor of K = 3.3 compared to the case of transmitting an unfragmented packet of initial length.