We study the energy consumption of wireless sensor network links applying three widely used digital modulation schemes, i.e. MQAM, MPSK, and noncoherent MFSK. For MQAM and MPSK, pulse shaping is considered. Both transmitted signal power and circuit power are taken into account where the former is modeled using the relationship between cutoff rate and signal-to-noise ratio (SNR) and the latter is modeled considering the power consumption of typical hardware used in wireless sensor nodes. For each modulation type, the optimum parameters for minimizing the energy per information bit are derived. Moreover, a comparison among the three modulation types is presented.
I. INTRODUCTIONEnergy efficient schemes are of paramount importance for wireless sensor networks (WSN) composed of battery-powered nodes, where energy efficiency directly affects the lifetime of the network.Energy consumption in energy-constrained networks has already been addressed in the literature. For example, Cui et al.[1] analyzed both transmission time and constellation size optimization for MQAM and MFSK modulations (both coded and uncoded), considering both transmission and circuit energy consumption. It has been shown that optimizing transmission time or, equivalently, constellation size, minimizes the total energy used for transmission of information.In [2], the energy performances of uncoded MPSK, MQAM, and MFSK modulations in both AWGN and Rayleigh fading channels were evaluated and compared for very short-range (less than 10 meters) communication. It has been concluded that MQAM is more energy-efficient than the other modulation schemes in this case.More recently, the combination of Reed-Solomon codes with MPSK, MQAM, and MFSK modulations has been studied in [3], where an optimum pair of coding and modulation were derived by computing required energy per bit.In our previous work [4,5], based on the relations between capacity (and cutoff rate) and SNR of MQAM, MPSK, and noncoherent MFSK, the optimum transmission scheme (constellation size and information rate) that results in minimum energy was derived for distances up to one hundred meters between WSN nodes. In [5], a comparison among the three cited modulation schemes was performed. A fixed circuit power consumption model developed in [1] was used.