Towards a Multihomed GSM Gateway Richard Withnell, Dr. Christopher Edwards School of Computing and Communications, InfoLab21, Lancaster University, Lancaster, LA1 4WA, United Kingdom Email: { r.withnell, c.edwards }@lancaster.ac.uk Abstract—Multihomed devices are becoming increasingly com- mon, especially in the mobile domain. The majority of devices only allow the use of one network interface at a time; removing this limit would allow for improved bandwidth and reliability. An introductory multihomed cellular device, termed the GSM Scanner has been developed using basic networking concepts to support multiple interfaces. This device has been used to test multiple cellular networks to justify the enhancement of the GSM Scanner to a true multihomed gateway. Technologies and implementations, relating to using multiple interfaces in this context have been investigated in order to identify the key areas of intended research. I. I NTRODUCTION Due to the fragmented nature of wireless technologies and the number of different providers for these services, it is important for consumer devices to support a number of dif- ferent access mediums. Many mobile devices support multiple wireless technologies such as UMTS, WiFi and Bluetooth in order to provide ubiquitous Internet access. This model while increasing availability does not have any effect on the reliabil- ity or efficiency of the wireless links. This is due to using only one network interface at a time to provide Internet access. This is a simplistic approach that does take advantage of coverage levels in different locations, however it could also be improved. If more than one interface can be used simultaneously then the quality of service can potentially be enhanced. This is not a new idea, for example the PPP multilink protocol [1], proposed a method for using multiple logical data links in 1990. The fragmentation of current access medium makes this concept more complex due to potentially large differences in network behaviour between both technologies and providers. Mobility also has a large impact on the development of such ideas, while a large portion of research focuses on a single device communicating via multiple interfaces, this may not always be the case. For example in the context of NEMO [2] or MANET [3] , the gateway router would ideally be able to utilise multiple interfaces to improve the quality of service to all hosts on the network. In order to address some of these issues, an initial device known as the GSM Scanner has been implemented. This has been designed in order to take advantage of the differences in cellular coverage and network quality between providers. The current iteration allows for multiple cellular networks to be monitored simultaneously. The GSM Scanner has been used in an experiment across the Lancaster University campus to determine the differences between cellular providers. This is in order to justify the use of multiple cellular modems to improve reliability and efficiency of a devices network capability. The GSM Scanner provides a good basis for the implementation and future investigation of other multihomed technologies, specifically targeting the mobile network domain. II. BACKGROUND The concept of WarDriving first detailed by [4], is a method of searching for and collecting data on WiFi networks while in a moving vehicle. This concept can be used with malicious intent to specifically search for and then illegally access unsecured wireless networks; it can however also be used with GPS information to map WiFi availability and signal quality. Additionally it can be used for WiFi based localisation, using information such as the access point SSID’s and signal strength to approximate position. This can augment GPS positioning or replace it entirely in indoor scenarios. The act of WarDriving can also be applied in the context of cellular networks. While the access medium differs, the information retrieved remains the same, requiring the Cell ID of the cellular networks base station, signal strength and GPS coordinates. Combining and mapping cellular and WiFi cover- age via WarDriving can help to produce a more comprehensive view as to the availability of Internet access. The mapping of cellular networks is usually based on the modems received signal strength. While this can give a good indication as to the quality of the cellular link, it does not show the quality of the back haul. This can be achieved using standard network monitoring techniques. Combining network metrics with the signal strength and location data can provide a detailed view regarding the capability of cellular providers over a given area. A large scale cellular network measurement was performed in [5] using a wide range of metrics such as throughput, round trip time and DNS lookup time; which provides a detailed view of the performance of multiple cellular networks. For example, an unexpected relationship between TCP round trip time and the TCP retransmission rate was identified. This showed that networks with consistently low round trip times were more likely to drop packets, the explanation being that packets are more readily dropped from queues to keep the latency lower on these networks. ISBN: 978-1-902560-26-7 © 2012 PGNet