5.5 Wireless Networks and Challenges

Before we study the application of streaming video to wireless networks, it is conducive to gain a historic perspective on the wireless industry. The cellular concept was first conceived and developed in the late 1970s. When the first wireless systems, the Advanced Mobile Phone System (AMPS) and its variations, were deployed in the early 1980s, they were built strictly for voice communications. Generally, these analog cellular networks were considered as the first-generation (1G) wireless technologies. The advent of voice coding and digital modulation technologies brought the evolution to the second generation or 2G wireless networks. The leading technologies included Global Systems for Mobile (GSM) in Europe, the IS-95 CDMA and IS-136 in the United States, and Pacific Digital Cellular (PDC) in Japan. Similar to the 1G wireless networks, the 2G networks are mostly used for low data rate, circuit-switched voice applications.

In the past few years, the explosion of Internet traffic has inevitably increased the need for packet-based wireless networks. As a result, the circuit-switched 1G and 2G wireless networks have gradually evolved into packet-switched 2.5G technologies such as GPRS, EDGE, and 1X-EVDV to provide packet data services and further improve voice capacity, which will eventually be phased out by the 3G wireless technologies. Employing increased spectrum, highly sophisticated air interfaces, and packet switching at the core, the 3G wireless networks further improve the capability to provide advanced data services. The high data rate (up to 2 Mbps) provided by 3G networks is much higher than that of today's wireline networks. In addition, 3G technologies provide seamless roaming across global networks. With these advantages, the 3G networks can support a wide variety of data services, including real-time, streaming multimedia and fast Internet access. In the end, the evolution of the 3G networks will bridge the gap between the wireline and the wireless worlds.

Given that the Internet traffic increases dramatically and users desire ubiquitous Internet access, the next generation of networking systems will be data-centric with the addressed mobility consideration. IP-based communications systems, which enable much-higher data rates and network flexibility, will gradually predominate over the traditional circuit-switched systems. In recent years, enormous effort has been made to support IP in wireless networks. Protocols and programming languages, including WAP, WML, and J2ME, have been developed to adapt Web content to the limitations of handheld devices by reducing the amount of transmitted data with minimum sacrifice of information. Mobile IP networks have been designed to maintain consistent transport-layer quality as the remote terminal is constantly in motion. However, in developing IP-based wireless data networks, significant difficulties remain to be addressed. They are summarized next.

5.5.1 Dynamic Link Characteristics

The process of a mobile device transmitting and receiving radio signals through the air makes wireless transmission vulnerable to noise and interference. The shadowing effect, multipath fading, and interference from the other devices make channel conditions vary unpredictably over time. Changing the transmission rate as the channel varies does improve efficiency but results in data rate oscillation. Furthermore, mobility introduces difficulty in channel estimation and prediction, thus raises error rate. Two approaches have commonly been used to address this problem. The first approach employs sophisticated channel coding and interleaving technologies. For example, turbo coding, despite its complexity, is now standard channel coding technique in 3G UMTS. ^[27] This approach, however, heavily relies on the quality of channel estimation. The second approach, the link layer ARQ mechanism, performs error control by retransmitting lost frames. ^[28] Although insensitive to the quality of channel estimation, this approach introduces latency and delay jitters to IP packet flow. The trade-off between latency and reliability depends on the ARQ persistence, which defines the willingness of the protocol to retransmit lost frames to ensure reliable transmission. ^[29] The persistence can be expressed in terms of time or the maximum number of retransmissions.

5.5.2 Asymmetric Data Rate

A mobile terminal has limited power so that the uplink data rate is usually less than the downlink data rate. This limitation is less stringent because most data applications are asymmetric.

5.5.3 Resource Contention

As in wireline networks, users share channel resources in wireless networks. When multiple users run a variety of applications, the most salient issue is the significant variability in terms of QoS requirements such as error rate, latency, and bandwidth. The resource contention problem is already quite challenging in wireline networks. As the result of mobility and unpredictable link variation, dynamic network topology makes wireless networks even harder to coordinate. The Medium Access Control (MAC) layer uses a scheduler to determine the next user to be served based on an individual user's channel condition and QoS requirement. ^[30], ^[31], ^[32] Currently, this scheduler is developed only for downlink transmission because only the base station gathers all the user information. The uplink transmission is typically made through contention, yielding high delay jitters.

Overall, high transmission errors and variable latency are the major causes of data loss in wireless networks. In the past, IP-based data applications have been designed mostly for wireline networks, where links and subnetworks normally have relatively stable transmission rates at low error rates. Data loss is primarily due to network congestion and buffer exhaustion. As described earlier in this chapter, many techniques have been developed to support efficient packet transmission over wire-line networks. Unfortunately, they are not applicable to wireless networks. For example, in wireline networks, adding bandwidth can solve latency problems because bandwidth is not a paramount concern. However, in a wireless environment, this is quite difficult due to adverse channel condition and limited battery life of the mobile device.

^[27]3rd Generation Partnership Project, Technical specification group radio access network, physical layer aspects of UTRA high speed downlink packet access (Release 2000), 3G Technical report (TR) 25.848.

^[28]Wang, Y. and Lin, S., A modified selective-repeat type-II hybrid ARQ system and its performance analysis, IEEE Trans. Commun., COM31, 593–608, 1983.

^[29]Advice to link designers on link Automatic Repeat reQuest (ARQ), Internet Draft, March 2002, draft-ietf-pilc-link-arq-issues-04.txt.

^[30]Jalali, A., Padovani, R., and Pankaj, R., Data throughput of CDMA-HDR: a high efficiency high data rate personal communication wireless system, Proc. IEEE Vehicular Technology Conference, Tokyo, Japan, May 2000.

^[31]Andrews, M. et al., Providing quality of service over a shared wireless link, IEEE Communications Magazine, 39 (2), 150–154, 2001.

^[32]Tse, D., Forward link multiuser diversity through rate adaptation and scheduling, Bell Labs presentation, New Jersey, 1999.