3.5 How Secure are Wireless Devices?

Internet security, as many have seen it applied to corporate networks today, can be difficult to implement on wireless phones and PDAs for a variety of reasons. Most of these devices have limited CPUs, memory, bandwidth, and storage abilities. As a result, many have disappointingly slow and limited computing power. Robust security features that can take less than a second to process on a typical workstation can take potentially many minutes on a wireless device, making them impractical or inconvenient for the mobile user. Because many of these devices have merely a fraction of the hardware capabilities found on typical workstations, the security features on portable devices are often lightweight or even nonexistent from an Internet security perspective. However, these same devices are now being used to log onto sensitive corporate intranets, or to conduct mobile commerce and banking. Although these wireless devices are smaller in every way, their security needs are just as significant as before. It would be a mistake for corporate IT and information security departments to ignore these devices as they start to populate the corporate network. After all, these devices do not discriminate; they can be designed to tap into the same corporate assets as any other node on a network. Some of the security aspects as they relate to these devices are examined here.

3.5.1 Authentication

The process of authenticating wireless phone users has gone through many years of implementation and evolution. It is probably one of the most reliable security features digital cell phones have today, given the many years of experience service providers have had in trying to reduce the theft of wireless services. Because the service providers have a vested interest in knowing who to charge for the use of their services, authenticating the mobile user is of utmost importance.

As previously mentioned, GSM phones use SIM cards or chips that contain authentication information about the user. SIMs typically carry authentication and encryption keys, authentication algorithms, identification information, phone numbers belonging to the subscriber, etc. They allow users to authenticate to their own phones and to the phone network to which they are subscribed.

In North America, TDMA and CDMA phones use a similarly complex method of authentication as in GSM. Like GSM, the process incorporates keys, Authentication Centers, and challenge-response techniques. However, because TDMA and CDMA phones do not generally use removable SIM cards or chips, these phones rely instead on the authentication information embedded into the handset. The user's identity is therefore tied to the single mobile phone itself.

The obvious drawback is that for authentication purposes, TDMA and CDMA phones offer less flexibility when compared to GSM phones. To deploy a new authentication feature with a GSM phone, in many cases, all that is needed is to update the SIM card or chip. On the other hand, with TDMA and CDMA, deploying new authentication features would probably require users to buy new cell phones — a more expensive way to go. Because it is easier to update a removable chip than an entire cell phone, it is likely that one will find more security features and innovations being offered for GSM.

It is important to note, however, that this form of authentication does not necessarily apply to Internet-related transactions. It merely authenticates the mobile user to the service provider's phone network, which is only one part of the transmission if one is talking about Internet transactions. For securing end-to-end Internet transactions, mobile users still need to authenticate the Internet Web servers they are connecting to, to verify that indeed the servers are legitimate. Likewise, the Internet Web servers need to authenticate the mobile users that are connecting to it, to verify that they are legitimate users and not impostors. The wireless service providers, however, are seldom involved in providing full end-to-end authentication service, from mobile phone to Internet Web site. That responsibility usually falls to the owners of the Internet Web servers and applications.

Several methods for providing end-to-end authentication are being tried today at the application level. Most secure mobile commerce applications are using IDs and passwords, an old standby, which of course has its limitations because it provides only single-factor authentication. Other organizations are experimenting with GSM SIMs by adding additional security ingredients such as public/private key pairs, digital certificates, and other public key infrastructure (PKI) components into the SIMs. However, because the use of digital certificates can be process intensive, cell phones and handheld devices typically use lightweight versions of these security components. To accommodate the smaller processors in wireless devices, the digital certificates and their associated public keys may be smaller or weaker than those typically deployed on desktop Web browsers, depending on the resources available on the wireless device.

Additionally, other organizations are experimenting with using elliptic-curve cryptography (ECC) for authentication, digital certificates, and public key encryption on the wireless devices. ECC is an ideal tool for mobile devices because it can offer strong encryption capabilities, but requires less computing resources than other popular forms of public key encryption. Certicom is one of the main pioneers incorporating ECC for use on wireless devices.

As more and more developments take place with wireless Internet authentication, it becomes clear that, in time, these Internet mobile devices will become full-fledged authentication devices, much like tokens, smart cards, and bank ATM cards. If users begin conducting Internet commerce using these enhanced mobile devices, securing those devices themselves from loss or theft now becomes a priority. With identity information embedded into the devices or the removable SIMs, losing these could mean that an impostor can now conduct electronic commerce transactions using that stolen identity. With a mobile device, the user, of course, plays the biggest role in maintaining its overall security. Losing a cell phone that has Internet access and an embedded public/private key pair can be potentially as disastrous as losing a bank ATM card with its associated PIN written on it, or worse. If a user loses such a device, contacting the service provider immediately about the loss and suspending its use is a must.

3.5.2 Confidentiality

Preserving confidentiality on wireless devices poses several interesting challenges. Typically, when one accesses a Web site with a browser and enters a password to gain entry, the password one types is masked with asterisks or some other placeholder to prevent others from seeing the actual password on one's screen. With cell phones and handheld devices, masking the password could create problems during typing. With cell phones, letters are often entered using the numeric keypad, a method that is cumbersome and tedious for many users. For example, to type the letter "R," one must press the number 7 key three times to get to the right letter. If the result is masked, it is not clear to the user what letter was actually submitted. Because of this inconvenience, some mobile Internet applications do away with masking so that the entire password is displayed on the screen in the original letters. Other applications initially display each letter of the password for a few seconds as they are being entered, before masking each with a placeholder afterward. This gives the user some positive indication that the correct letters were indeed entered, while still preserving the need to mask the password on the device's screen for privacy. The latter approach is probably the more sensible of the two, and should be the one that application designers adopt.

Another challenge to preserving confidentiality is making sure that confidential information such as passwords and credit card numbers are purged from the mobile device's memory after they are used. Many times, such sensitive information is stored as variables by the wireless Internet application and subsequently cached in the memory of the device. There have been documented cases in which credit card numbers left in the memory of cell phones were reusable by other people who borrowed the same phones to access the same sites. Once again, the application designers are the chief architects in preserving the confidentiality here. It is important that programmers design an application to clear the mobile device's memory of sensitive information when the user finishes using that application. Although leaving such information in the memory of the device may spare the user of having to reenter it the next time, it is as risky as writing the associated PIN or password on a bank ATM card itself.

Yet another challenge in preserving confidentiality is making sure that sensitive information is kept private as it travels from the wireless device to its destination on the Internet, and back. Traditionally, for the wired Internet, most Web sites use Secure Sockets Layer (SSL) or its successor, Transport Layer Security (TLS), to encrypt the entire path end-to-end, from the client to the Web server. However, many wireless devices, particularly cell phones, lack the computing power and bandwidth to run SSL efficiently. One of the main components of SSL is RSA public key encryption. Depending on the encryption strength applied at the Web site, this form of public key encryption can be processor and bandwidth intensive, and can tax the mobile device to the point where the communication session itself becomes too slow to be practical.

Instead, wireless Internet applications that are developed using the Wireless Application Protocol (WAP) use a combination of security protocols. Secure WAP applications use both SSL and WTLS (Wireless Transport Layer Security) to protect different segments of a secure transmission. Typically, SSL protects the wired portion of the connection and WTLS primarily protects the wireless portion. Both are needed to provide the equivalent of end-to-end encryption.

WTLS is similar to SSL in operation. However, although WTLS can support either RSA or ECC, ECC is probably preferred because it provides strong encryption capabilities but is more compact and faster than RSA.

WTLS has other differences from SSL as well. WTLS is built to provide encryption services for a slower and less resource-intensive environment, whereas SSL could tax such an environment. This is because SSL encryption requires a reliable transport protocol, particularly TCP (Transmission Control Protocol, a part of TCP/IP). TCP provides error detection, communication acknowledgments, and retransmission features to ensure reliable network connections back and forth. But because of these features, TCP requires more bandwidth and resources than what typical wireless connections and devices can provide. Most mobile connections today are low bandwidth and slow, and not designed to handle the constant, back-and-forth error-detection traffic that TCP creates.

Realizing these limitations, the WAP Forum, the group responsible for putting together the standards for WAP, designed a supplementary protocol stack that is more suitable for the wireless environment. Because this environment typically has low connection speeds, low reliability, and low bandwidth in order to compensate, the protocol stack uses compressed binary data sessions and is more tolerant of intermittent coverage. The WAP protocol stack resides in layers 4, 5, 6, and 7 of the OSI reference model. The WAP protocol stack works with UDP (User Datagram Protocol) for IP-based networks and WDP (Wireless Datagram Protocol) for non-IP networks. WTLS, which is the security protocol from the WAP protocol stack, can be used to protect UDP or WDP traffic in the wireless environment.

Because of the differences between WTLS and SSL, as well as the different underlying environments that they work within, an intermediary device such as a WAP gateway is needed to translate the traffic going from one environment into the next. The WAP gateway is discussed in more detail in the infrastructure section of this chapter.

3.5.3 Malicious Code and Viruses

The number of security attacks on wireless devices has been small compared to the many attacks against workstations and servers. This is due in part to the very simple fact that most mobile devices, particularly cell phones, lack sufficient processors, memory, or storage that malicious code and viruses can exploit. For example, a popular method for spreading viruses today is by hiding them in file attachments to e-mail. However, many mobile devices, particularly cell phones, lack the ability to store or open e-mail attachments. This makes mobile devices relatively unattractive as targets because the damage potential is relatively small.

However, mobile devices are still vulnerable to attack and will become increasingly more so as they evolve with greater computing, memory, and storage capabilities. With greater speeds, faster downloading abilities, and better processing, mobile devices can soon become the equivalent of today's workstations, with all their exploitable vulnerabilities. As of the writing of this chapter, cell phone manufacturers were already announcing that the next generation of mobile phones will support languages such as Java so that users can download organizers, calculators, and games to their Web-enabled phones. However, on the negative side, this also opens up more opportunities for users to unwittingly download malicious programs (or "malware"). The following adage applies to mobile devices: "The more brains they have, the more attractive they become as targets."