We argue that the more sophisticated and convenience-oriented requirements of motorists in an affluent economy will drive development of the networked car towards open systems and eventually convergence with the Internet.
In an emerging economy there are typically very specific contextual stumbling blocks. Violent crime and its specific local nature is a case in point for South Africa. By contrast, then, the network(s) will evolve in a manner to best satisfy the single most compelling need for a networked car, and might favor proprietary solutions.
We use the United States as an example of an evolved, affluent market for connected cars, whereas the potential market in South Africa, an "advanced" developing country, has a very clear problem: pervasive violent crime.
The paper describes the South African situation and contrasts some aspects with the situation in the United States.
We argue that the driving forces that influence the establishment of platform and network architecture for connected cars in South Africa are significantly different from those at work in the United States. We generalize some of these drivers and highlight the importance of lobbying motor car manufacturers, network infrastructure providers, and value-added service providers so that there is medium- and long-term benefit for all in a more open platform.
This paper does not aim to propose detailed implementations of a network of connected cars, or the various technical issues that affect the choice of such architectures. In the interest of clarity, we briefly outline our concept of a connected car network:
Although many of these issues hold not only for passenger cars, but also for commercial vehicles, we concentrate on the passenger car segment as a market for connected cars. The requirements of operators of commercial vehicles in the road transport industry are very specialized and have in many ways been addressed by custom solutions. These include trunked radio for voice and data communication, and proprietary satellite and/or terrestrial network tracking.
In our discussion of connected cars, we distinguish between the following contributors.
This segmentation is somewhat arbitrary, and clearly one entity can act in more than one of these roles.
As the basis for the ensuing arguments, rather than trying to provide a prioritized or exhaustive list, we mention a few reasons why consumers would be interested in connected cars:
For a recent survey of available services, see .
We assume that a connected car platform and infrastructure will be established only if there is a sufficient business case. We assume further that participants -- be they motor manufacturers, terminal equipment providers, network infrastructure providers, or value-added service providers -- will get involved only to the extent that they see a sufficient likelihood of profitable business. Likewise consumers will pay for a service only if they believe that they derive sufficient value from their expenditure.
As in many other choices and options facing South Africans, personal safety and security will be a major and possibly dominant consideration when acquiring network services for their cars.
A significant problem that South African motorists face is the extent of violent car hijacking, or carjacking. Since 1996 the South Africa Police Service separately keeps statistics for carjacking, as a special case of robbery with aggravating circumstances. In 1999 the national incidence ratio of carjacking was 18 per 100,000 population per year. In the densely populated Gauteng province it was as high as 60 incidences per 100,000 population per year, and about 10% up from the 1996 figure.
To add to the frustration of motorists, (ordinary) theft of motor vehicles occurs at about six times the rate of carjacking. During 1999, more than 50,000 vehicles were reported stolen in South Africa. To place this in perspective, for every five new cars sold, one car was stolen.
As a result, and under pressure from insurance companies, motorists are required to fit their cars with gear locks and electronic immobilizers with anti-hijacking features. Discounted premiums are offered for cars with a transponder-based tracking device fitted, and for more expensive cars insurance companies require these. In summary, motorists are used to spending additional money on devices and services to improve the security of their motoring experience.
The security industry is possibly not the poster child for business ethics, and lock-in is a popular way to limit customer churn. There are currently three major vehicle-tracking companies operating in South Africa. In addition to the monthly service fee, the tracking services require proprietary hardware at a cost of between 12 and 24 months' subscription to the service, depending on model selected. Should the service somehow not be satisfactory, the cost of changing providers is substantial and often not worth it.
A similar situation occurs in the domestic security industry where domestic burglar alarms are linked to armed response centers via a proprietary radio interface. The services of the armed response team can be secured at a monthly fee, but changing response providers requires new communication hardware.
The proprietary transponder tracking system can be replaced by a service based on GPS and an open standards wireless network. This service would additionally have panic button activation, where the transponder-based services require a voice call to a control center to activate the transponder. If motorists are offered this proposition, and additionally it can provide convenient services like navigation and travel tips, it is likely to prove to be quite attractive.
South African consumers are unusually docile by U.S. standards. In most industries -- including even retail and hospitality -- service levels are remarkably low, but tolerated by most consumers. The motor industry in South Africa has traditionally been a "seller's market" in that waiting times for popular models often tend to be long, and the availability of a prospective buyer's desired configuration and options is often considered a luxury.
On this premise, we suggest that motor industry consumers will not raise their voices in objection to less desirable connected car network platforms and architectures, as long as they are presented with a reasonable bouquet of services. This view is reinforced by the fact that at least three proprietary transponder-based satellite/terrestrial network tracking services exist, apparently profitably, in South Africa.
By contrast the United States, largely as a result of the widespread (relative) wealth, has evolved a highly consumer-friendly business culture. Healthy competition exists in a large market, which forces industry to entice customers with more attractive value propositions, rather than being the first or only agency to provide a service or product.
The total sales of passenger and light commercial vehicles for all South African manufacturers for 1999 amounted to 285,000 units. By comparison, in 1999 General Motors (GM) sold 5.2 million cars and trucks in the United States only. With GM claiming a 29% market share in this segment, that leaves annual sales of passenger and light commercial vehicles in South Africa at 1.5% of that in the United States. Of the total passenger and light commercial vehicle market, luxury vehicles comprise only about 3%.
South Africa has a notable motor manufacturing industry. However, for most manufacturers the vast majority of research and development takes place at the headquarters or the designated research and development center of the manufacturer. This removes the freedom to design an optimal connected car network from the local industry, as the terminal equipment is likely to be a given.
In South Africa, for the near future at least, a network interface will only be found in mid- and upper-class luxury models. Many of these models are fully imported, with no local assembly. No modifications to the vehicles are allowed, as this would require all service processes and documentation as well as parts stocking processes to be revised. The tiny market makes this non-feasible. As a result these cars will have whatever systems are fitted in their country of origin. In South Africa, for the majority of luxury cars, this will imply European and especially German-flavored systems. It is not entirely unlikely that such a system would be GSM-based on the link layer, since most German models offer an integrated GSM phone -- the majority with data capabilities -- as an option already.
In South Africa then, the German cars will probably pioneer the local connected car network. BMW South Africa already offers two levels of CD-ROM-based navigation system as an option in all its new models, and an integrated GSM telephone as an option in most. As the first to market, the foundation for the network of connected cars in South Africa will likely be laid by the German motor industry.
Given the relatively small initial market as argued above, and that the market is still divided between a few motor manufacturers, it is highly unlikely that a motor manufacturer in South Africa will run its own network in the same way GM runs its OnStar network. ,  Because quite a few manufacturers provide a maintenance plan and roadside assistance already, it is conceivable that they may wish to provide these services themselves. One valid reason could be because they want to differentiate on after-sales service and support, and not outsource such a function.
Figure 1 shows a crude representation of some of the elements of a family of connected car network implementations. The figure serves as an illustration of the arguments, and is not an attempt to exhaust the various options.
Assume that a network can be implemented using the protocols and services along a vertical slice anywhere along the diagram. We will discuss a few examples, with some of their implications and assumptions.
Figure 1: A simplified representation of classes of connected car networks
In South Africa, it is highly likely that the physical and link layer will be provided by an existing network such as the two -- and soon to be three -- GSM networks. A less likely contender may also be the trunking radio networks, but the coverage, equipment availability, and established infrastructure for billing and customer support of the GSM networks favor them strongly. In current network infrastructure this would be via circuit-switched GSM data (GSM CS), with a strong likelihood of migration to general packet radio service (GPRS) once it is deployed.
In the currently available infrastructure it is therefore possible to implement a connected car network at A in Figure 1. This would be the equivalent of a bulletin board system, where the network provides a 9.6 or 14.4 kbps circuit to a proprietary system hosted at a GSM network switch. If a variety of services and mobile network roaming are desired, this is clearly not a desirable solution. If personal security applications are the main consumer requirement, it is not unlikely that manufacturers may team up with the mobile GSM networks to host vehicle tracking and emergency response services in this way.
The equivalent for the United States (or other countries) lie to the right of E in Figure 1.
For South Africa (and other countries with GSM networks) a network between B and C provides the best of all worlds to consumers and service providers alike. Even Internet Protocol (IP) over GSM without higher-level protocols could be workable. IP Version 4 (IPv4) should be sufficient if no roaming beyond the home GSM network is required. (Mobility issues are then managed on the GSM level, and the car is effectively a portable and not mobile node on the IP level.) Mobile IP will add the option of roaming on other GSM networks. It is not clear whether this will add significant value in the South African context, as the existing two networks do not allow roaming on each other's networks, and neighboring countries in the region, although some do have GSM networks, are unlikely to provide services for cars.
Using IP would shield value-added service developers from having to update their applications every time an improved link layer service becomes available (such as when GPRS supersedes circuit-switched GSM data.) Likewise consumers will have the choice of subscribing to any service that can be delivered over IP, and do not have to be limited to the services provided by their own GSM network.
Again a similar situation exists between D and E for non-GSM (for example, CDMA [Code Division Multiple Access]) mobile networks elsewhere such as in the United States.
The area between C and D in Figure 1 represents the likely future scenario, where current mobile cellular networks are replaced by third-generation (3G) or UMTS (Universal Mobile Telecommunications System) networks. This need not affect the availability of services if IP and higher-level protocols were also used.
Additional benefit, especially for prospective value-added service suppliers, will accrue if the connected car network also embraces some higher-level open protocols such as Transmission Control Protocol (TCP) or User Datagram Protocol (UDP), and others.
A network at F is feasible only if the market size warrants the roll-out of an entirely separate network. This is unlikely to ever be the case in South Africa. However, in the United States there are several such networks -- like Mobitex networks -- that serve wireless data customers only. (The 3COM Palm VII Palm.Net is an example of such a network.)
South Africans have accepted the reality of having to spend money on personal safety and security. Services that enhance the personal safety and security of the motoring experience are likely to be the dominant requirement of South African connected car consumers. These services are likely to be commercially successful, especially since additional features like relevant traffic information and on-board navigation will encourage consumers to spend money on the new technology.
Research and development decisions on connected car network architectures particularly of German car manufacturers are likely to be imported and adopted in South Africa.
Given the significant investment in infrastructure and, subsequently, sufficient nationwide coverage, the network of connected cars is highly likely to be built on top of South Africa's GSM networks. This will also allow the connected car network to evolve and improve along with the growth of the GSM networks.
Technical issues of protocol on various levels are unlikely to significantly affect the architecture and platform for a network of connected cars in South Africa. Main drivers will be business decisions by service providers (lock-in versus an open competitive approach) as well as partnering decisions made by motor manufactures: whether to partner with network infrastructure providers or information and application service providers.
Various references were used to extract relevant market statistics and segmentation; these are not cited here but can be obtained from the author.
 Ratliff, Evan, " The Telematics Options", Wired 7.07, http://www.wired.com/wired/archive/7.07/options.html
 Lappin, Todd, "The New Road Rage", Wired 7.07, http://www.wired.com/wired/archive/7.07/gm.html
 Welcome to OnStar, http://www.onstar.com/
Wessel du Preez is manager and a technical evangelist in the Wireless Systems and Applications group at CSIR (Council for Scientific and Industrial Research) Mikomtek. His professional interests include stratospheric telecommunication and networked cars. He lives and works in Pretoria, South Africa.