Android Calling

Last November, Google and 33 other companies announced the Open Handset Alliance (OHA), a new industry consortium that, in promoting open standards for mobile devices, promises to reinvent the cell phone–and possibly the entire wireless-telecommunications industry. While that’s a tall order, I suspect that handsets, which Google intends to make as customizable as laptop computers, are just the beginning of the company’s mobile efforts.

The word “open” in OHA’s name is not just a buzzword: it signifies a radical ­departure from today’s cellular networks, especially those in the United States. Today’s cellular ecologies aren’t exactly closed; it’s possible to load third-party applications onto some cell phones, and websites belonging to third-party providers such as Google can be accessed. Verizon (not an OHA member) has even announced its willingness to open its network to non-Verizon phones. But that openness is all at the periphery: wireless providers today offer just enough choice in phones, features, and services to remain competitive while preventing consumers from using rival tech­nology and defecting to other carriers.

Indeed, companies like Verizon and AT&T (also not an OHA member) operate vertically integrated telecommunication ecologies of stores, resellers, content providers, and network services–all with the goal of extracting as much revenue as possible from a customer base that’s kept captive with multiuser contracts, exclusive hardware offerings, and free in-network calling. The wireless industry’s favorite metric is “average revenue per user.”

This approach–using integrated services to extract maximum revenue–will slowly evaporate if OHA is successful. And the key to that success will be Android, a new software “stack” for mobile phones that’s based on open-source software and a revolutionary programming paradigm.

Android is called a “stack” because its software extends from the lowest levels controlling the phone’s hardware to the highest levels of user interaction. At the bottom is a stripped-down version of the Linux kernel (the heart of the Linux operating system). On top of the kernel is the open-source WebKit Web browser (also used by Apple’s iPhone) and several other open-source programs. On top of this are user applications that are beautiful but, at least in the developer’s preview version, primitive and buggy. Once Google finishes this release, Android is going to look as pretty as the iPhone–and it will be just as functional, if not more so.

The first thing to point out is that you can’t run Android on a phone that you might have today. Instead, manufacturers like HTC, LG, Samsung, and Motorola (all OHA members) will need to adapt it to future handsets. If the consortium has its way, these phones will be available in stores in the second half of 2008. You’ll also be able to buy an Android-based phone over the Internet and drop in a chip from the cell phone that’s in your pocket today, assuming you have a cell phone from T-Mobile, AT&T, or another provider that uses the GSM transmission standard.

Android’s developers envision a world where today’s integrated wireless systems are reduced to a set of relationships between parts that are more or less interchangeable. Consumers will be free to load their phones with applications of their own choosing–free applications, applications available for sale, and custom applications developed by enterprises for their employees. These applications will be able to communicate with third-party services offered over the Internet–using any available communications pipe, be it the cellular network, a nearby Wi-Fi connection, or even a Bluetooth connection from another phone.

This feature is good news for companies whose business model is based on providing information to consumers–and possibly showing them advertisements. It’s terrible news for those whose business model is based on charging consumers to access hardware built into their own devices–the way Verizon does with its VZ Navigator, a navigation service that uses the GPS receiver standard in new ­Verizon phones and costs $9.99 a month or $2.99 a day.

An Architecture for Mashups
Does Android have what it takes to blow apart the silos in today’s cellular-telephone marketplace and open it to the chaos of competition? Although the developer’s preview is flawed, Android’s architectural choices and development tools show great promise.

What’s more, though Android itself is brand new, it draws from proven Google software. Consider Google Maps, which has been one of Google’s most successful offerings to date. With little effort, practically any Web developer can create a “mashup” that drops geographical annotations like photos, routes, and notes about particular places onto a beautiful, interactive, and highly functional map (see “Second Earth,” July/August 2007). Android might, because of the way it is built, be able to replicate the versatility of Google Maps with respect to the phone itself. Unlike conventional mobile operating systems, which see each application as an essentially monolithic whole, Android splits a given app into multiple parts, with well-defined interfaces between them. This makes it easy for developers to write ­component-­based applications.

Interestingly, Android’s component-based structure could also extend battery life. On a system like Windows Mobile, programs spend a lot of time running in the background, where they use memory and drain the battery. With Android, only a tiny piece of each application should need to run at any time; the other parts could shut down. As a result, it should feel more responsive yet use less power.

Developing for cell phones is normally much harder than developing for desktops: because far fewer developers work with cell phones than with desktop applications, the tools are less polished and harder to use. With Android, Google has rewritten the rules here as well. Developers write Android applications in the ubiquitous Java programming language, so there are already millions of would-be Android programmers. The free Android developer kit includes a telephone emulator, which lets any developer with a PC, Mac, or Linux desktop write and test Android applications. The emulator even makes it possible to control the speed and quality of the simulated phone’s network connection, allowing developers to see how their programs will behave on phones in poor coverage areas without having to load the applications onto real phones.

If Android succeeds, it will have a major impact on wireless carriers. A phone running Google’s component-based operating system, after all, would treat wireless operators like Verizon and AT&T as just another way to reach data services on the Internet. Such a phone could turn today’s wireless providers into commodity data communications networks that also happen to carry voice. This would force the providers to compete in every area–network quality, handset quality, and price–without allowing good performance in one area to lock customers in and support ­mediocre performance in another. T-Mobile (an OHA member) already gets it: some of its newest phones allow calls over either T‑Mobile’s GSM network or Wi-Fi. I think this change will happen even without the Google phone. I’m seeing more unlocked phones with Wi-Fi capability from companies like Nokia (not an OHA member); just drop in an AT&T SIM card and they’ll play on AT&T’s network. But though the change may happen anyway, Google is pushing it along at a faster clip.

Just as Google’s place in the wireless world is a work in progress, so too is Android, which I suspect will not be limited to cell phones. If it’s successful, we’re likely to see Android as the basis of other handheld devices: digital cameras, GPS receivers, or even lightweight tablet computers. If Android really works, it’s going to change the face of mobile computing.

Simson Garfinkel is an associate professor of computer science at the Naval Postgraduate School in Monterey, CA, and a contributing editor at Technology Review.