For general irrigation you will find very good results using ET-based algorithms. Benchmarking has shown almost identical control solutions between calculated ET vs. measured with tensiometer class sensors. The installation cost and maintenance really make soil moisture measurement cost prohibitive for the average consumer. Use of low cost sensor and poor installation considerations make for poor control.
This year I am evaluating the Vegitronix moisture sensor which is a new technology. My test setup uses a WebControl module to provide the A/D conversion for the sensor, local temperature and humidity measurement and IP interface for data transfer. The attached shows the probe's performance to observe the soil moisture in a desert climate using a drip irrigation system.
Given that the only real variable in this climate and this test is the solar effects it can be seen that the voltage lows range from about 1.6 volts to 1.4 volts depending upon the temperature highs that range from about 80 to 110 for moisture loss over a 48 hour period. This means that is looks promising for a low cost way to measure measure for purpose of control. Control points would need to be determined, but the behavior of the sensor is relatively linear for the range of interest so this step would not be too difficult with a little experimentation.
Even though it is technically adequate, it still is a pain to install the instrumentation considering the electronics need to be protected from the elements and the results will be no better than that received from an ET calculation. Depending upon the degree of microclimate there is likely the ability to use measurements from neighborhood sources such as weatherunderground or AWS and regional data from Weather.com or NWS and not incur any instrumentation costs or maintenance.