PNW June 2011 meeting about cell phones was held in a phone museum, the Seattle Museum of Communications. Shown L-R is PNW vice chair Rick Chinn, Mary Mehrens, Micah Mehrens, Scott Mehrens of Motorola, Dan Mortensen, PNW Chair 2011-12.
At our June meeting, Scott Mehrens, software engineer with Motorola Mobility, spoke about the modern cell phone, and gave some insight into what makes these things tick.
Scott started with a brief history of Motorola, which began in 1928. Motorola made its name with car radios in the early 1930s. By the 1950s, non-cellular car phones were made, and by the 1960s, the space program used Moto radio links hinting at cellular networks that were demonstrated in the late 1970s. 1983 saw the first, huge analog cell phones, and we have seen the evolution to digital and greatly increased capabilities with less size and weight.
Next, Scott spoke on cell phone handset acoustics. His role is mainly DSP work, while others handle hardware/product engineering and acoustic testing. Speakerphone mode is differentiated from the handset mode and has different processing needs, and both have severe design constraints for packaging, size, cost, power and so on. He compared a graph of a phone test to a common studio monitor.
The microphones are very small MEMS (micro electro mechanical system) devices.
Speakerphone and earpiece transducers were discussed. Earpieces are especially dependent on position & coupling to ear; wideband audio is also coming (16k sampling vs. 8k). HATS (head and torso simulator) and artificial mouth acoustic tests were described.
An overview of phone signal processing was then reviewed. A lot of processing is done on a phone to allow comfortable conversation in many difficult situations. Speakerphone mode especially needs much processing for echo cancellation and noise suppression. Processing when both sides are talking (double talk) is hard. Sidetone (hearing yourself a little in your own speaker) is needed, as is "comfort noise" that adds some noise when things are quiet so you don't think you dropped the call. The details are generally proprietary, a manufacturer's "secret sauce," and a lot of work is done on the algorithms. Motorola gets about 15-30dB of echo reduction.
Phone acoustic testing was touched on, with (the standard) PESQ (perceptual evaluation of speech quality). Sometimes, it does not appear appropriate.
Scott is now working a lot on multimedia DSP for phones. He gave an overview of the Android "Gingerbread" OS multimedia audio framework and its audio effects like mixing, spectral enhancement, dynamics processing, and maximizing loudness. This is envisioned for gaming. Motorola expects things like 3D audio simulations and has a virtualizer with HRTF (head related transfer function) and other functions.
Other phone features may include spectral enhancement (EQ), personal EQ profiles and pre-call noise processing prediction. There are in fact, so many audio process that there is a big database for each phone model to optimize the sound.
A fixed point math review was next, and its relation to processing digital audio on a phone. Lastly, a general overview of the GSM phone system, and its myriad of acronyms, was given.
After the break, attendees were free to explore the Museum of Communications. Co-located in a Seattle working phone company exchange building, the museum holds the history of the telephone company and the central office switches. Exhibits range from 1876, starting with a model of Alexander Graham Bell's first attempt at creating a communications device, to the modern phones that we all take for granted today. Automated displays and special exhibits permitted hands on participation. Normally, a self guided tour is available as well as guided tours provided by the volunteers. There are 3 working telephone switches on display spanning the 20th century. Attendees could trace a call through the switch from the caller's phone going off-hook, dialing, and the ring at the distant end. http://www.museumofcommunications.org/
