RF Energy Harvesting System and Circuits for Charging of Wireless Devices Using Spectrum Sensing

1.1 Preprocessing in energy harvesting system

Spectrum is a scarce source. In wireless communication systems, efforts have been made to use frequency spectrum with policies and priorities in order to maximize the spectrum efficiency. The main idea here is to allocate empty spectrum holes over time, frequency and space to secondary users while the interference with primary user is minimum. Several approaches are proposed for spectrum sensing, such as energy detection [11, 12, 13], matched filter [11, 12, 14], cyclostationary detection [15, 16], spectrum sensing based on covariance matrix [17, 18] and wavelet based spectrum sensing [19]. By studying energy detection, it can be understood that this approach is based on detecting the signal power such a way that secondary users detect the signal power received from primary users. Then they compare it to some predefined threshold level and then decide whether they can use the primary frequency band or not. Well, here is the novel preprocessing idea which we are exploited in this chapter:

“In RF energy harvesting, we use RF signal power and convert it to DC power for charging batteries. On the other hand, energy detection algorithms give us the ability to detect the ambient signal power and compare it to a predefined threshold level. So, we do the same here as well. But with the goal of detecting a strong RF signal. In other words, by exploiting this preprocessing, we compare our received RF signal power and if it is greater than a threshold, then we will decide that frequency contains our required power for energy harvesting purposes and switch our circuit to that frequency. In the simulation process, this threshold is set to 0dBm which is a reasonable and practical assumption based on Figure 1.”

2.1 Battery model

For simulation stage and performance evaluation of our proposed system, we must be able to model the battery that we intend to charge. There are various battery models with different structures and complexities. Electrochemical models [20, 21, 22] are usually used for battery physical design, performance and power generation optimization. Mathematical models [23, 24, 25, 26], are much more effective. Random events for predicting battery systematic behaviors like battery life time and efficiency are discussed using mathematical equations. Electrical models [27, 28, 29, 30, 31, 32, 33] are placed somewhere between mathematical and chemical models in terms of accuracy and utilize the combination of voltage sources, resistors and capacitors.

For the goal of this chapter, an accurate and effective battery model is proposed based on battery model proposed in [34]. This model provides an easy extraction procedure, gives run time, static and transient responses and also contains all of the electrodynamic characteristics of the batteries. Figure 3 shows this proposed model.

3.1 Spectrum sensing simulations

We consider Mrandom base stations in the aforementioned frequency band. Using OFDM transmitter, the output signal is generated and by exploiting energy detection, our desired RF signal for energy harvesting circuit is obtained. Parameters of simulation are given in Table 1. Note that for digital to analog converter, a13th order Butterworth filter is used (normalized cut off frequency equals to 120).

RC pulse shaping time response is as follows [35].

where αand Tsare roll-off factor (takes values from 0to 1) and symbol repetition rate respectively.

As it can be seen in Figure 4, there are four signals available in the spectrum and in order to find the RF signal with maximum power, the area under each signal should be calculated i.e. its power. After applying energy detection we find that in this case, maximum power RF signal happened at 915MHz. Therefore, the filter is set to select this signal out of the spectrum.

After finding the frequency in which our desired RF signal exists, spectrum should be fed to a filter with the central frequency of 915MHz and the bandwidth of 10MHz. In Figure 5, filter characteristics in terms of frequency response is given. Note that this filter has the ability to be tuned to select the maximum frequency each time. Also, this filter should reject the rest of frequency band otherwise we face some challenges such as power loss and circuit design complexity.

Figure 6 shows power for filter, spectrum and the output of filter (Antenna input signal). We use OFDM transmitter as mentioned earlier with QPSK modulation.

3.2 Circuit design and simulation

In previous section, RF signal with maximum power is identified and extracted from spectrum. Now, this signal represents the input of our charging circuit. We assume a1mW signal with the frequency of 915MHz as our input to execute our simulation in ADS. First, we need to increase the DC level of our signal using a DC voltage multiplier circuit as it is shown in Figure 7. Note that we have a 4-stage and a 6-stage voltage multiplier in our proposed circuit. The connection between these two multipliers results in increasing of the output voltage.

In Figure 8, Efficiency of the rectifier circuit is shown versus different values of Pin. As it can be seen, when the input power is 0dbm, highest efficiency is obtained at the frequency of 915MHz.

Figure 9 indicates the return loss (s11) for different frequencies and minimum return loss is obtained at 915MHz frequency at a circuit input power of 0dbm. This is resulted from our designed matching circuit and shows its desirable performance.

In Figure 10, rectifier circuit output voltage and current are indicated with the maximum at 915MHz.

For frequencies of 580, 650, 760and 915MHz, the outputs of voltage multiplier circuit are shown in Figure 11. As it can be seen, we obtain a 8.8V DC voltage for our input signal.

Figure 12 shows the higher efficiency of our proposed circuit comparing to three other methods. That is because of exploiting a 10-stage voltage multiplier (a 4-stage connected to a 6-stage).

Complete charger circuit is proposed in Figure 13. Output currents for aforementioned four frequencies are given in Figure 14. Also, output voltages are indicated in Figure 15. So knowing that power equals to current times voltage, we obtain 532μW output power for 915MHz signal. It should be noted that voltage drop in 915MHz (8.8V to 3.6V) is because that by connecting battery to charger circuit, battery charging process starts and in this process power must be constant, so by increasing the current drawn by the battery, output voltage drops.

3.3 Antenna simulation

Receiver antenna is a broadband type with linearly polarized radiation. Considering that our bandwidth is 500MHz, according to previous experiences, we require an Ultra Wide Band (UWB) antenna (Figure 16). By means of simulation and measurement, reflection coefficients of our designed antenna are indicated in Figure 17. Note that because of the size of this antenna, we did not use it obtain fabrication results. A challenge may be rise to decrease the size of UWB antenna and improve our designed system performance in the future.

3.4 Charger circuit fabrication

Test results for our fabricated charger circuit are indicated in Figure 18. In terms of experimental results, we obtain 6.89V of output voltage at the frequency of 915MHz (which obtained 8.8V in simulation results), where our input signal is a 0dbm generated by R&S®SMB100A signal generator. For a 10kΩresistive load, load voltage is 2.12V and we can calculate the output power as 450μW.