Internet Via Satellite: The Challenges

Current technology allows the Internet interaction channel to remain within the ISDN link. Satellite  technologies are uniquely poised to address some of the key challenges that must be met for the Internet to continue its rapid expansion and advancement: specifically, high-speed access for rural and remote areas; the distribution and delivery of media-rich content; and tight integration with existing technologies. With advances in technology and the miniaturization of electronic components, bidirectional communications can be carried over satellite, eliminating the necessity of the subscribers ('end users’) terrestrial Internet connections that are inherent with the existing one-way satellite systems.

Internet Via Satellite

The Internet
User (subscriber) accesses the Internet via his/her ISP, which then establishes access to an Internet point of presence via a gateway station for connection to the satellite’s network access gateway (NAG). The NAG then sends the message across the proximate backbone network of the recipient’s ISP, which is finally delivered to the intended recipient’s address. For this type of application, channel capacity needs to be shared by different users. Subscribers’ access to satellite using a subscriber unit (e.g., handheld, pocket-sized phones) may be the simplest feature of satellite mobile systems of the future where the subscribers will be able to access voice, facsimile, data, and paging services with acceptable high quality. There are direct benefits for users accessing Internet via satellite, including Connection through or bypassing terrestrial networks and provision for direct-to-home services via the digital video broadcasting (DVB) IP. DVB technology uses multiplexing (e.g., MPEG-2) for data packeting but with different channel modulation. MPEG-2 is the broadcastquality compression standard for video communication. Provision for higher throughput and possibly asymmetric links. Allowance for multicasting transmission; that is, sending one message to authorized users instantaneously within the network through a variety of earth stations. Provision for faster access (or response) time and delivery of information to any inhospitable terrain or region where there is rapidly expanding customer demand but insufficient terrestrial infrastructure. Of course, there have been attempts to deliver IP via satellite, but the satellite technologies have focused on connection-oriented transmission protocols, which are really suited for voice traffic rather than IP, unnecessarily squandering
expensive capacity.

Technical Challenges
The TCP/IP shortcomings in the typical satellite environment include latency (degradations due to slow-start), data security over the satellite link, and optimization (window size, traffic acknowledgment for short transmissions over high-latency channels, as well as scheduling capacity). These shortcomings demand attention in order to provide more efficient, secure, and application-expansive internetworking via satellite.

Latency. Latency is a critical parameter of communication service quality, particularly for interactive communications and for many standard data protocols. TCP/IP performs poorly over high-latency or noisy channels. Bit error rates (BER) of 10^-7 can be acceptable in telephone environments, but this level of performance will render TCP/IP almost unusable. Geostationary satellite links are inherently high-latency and can be noisy. For example, if circuits are positioned in geosynchronous orbits, they suffer a transmission delay of about 119 msec between an earth terminal and the satellite, resulting in a user-to-user delay of 238 msec and an echo delay of 476 msec. Latency in voice communications becomes noticeable with a round-trip delay over 100 msec. To overcome the slow-start this delay causes, an enhanced communications protocol over the satellite link must be implemented which terminates and re-originates TCP connections at both ends of the satellite link without sacrificing the discovery process and still maintaining end-to-end standards compliance.

Optimization. Window size and scheduling capacity are optimization issues, which affect data throughput and network efficiency. Window sizing works simply by allowing a transmitter to send a number of packets before waiting for an acknowledgment from the receiver. Typically, the window size is short in comparison to the 476 msec delay of a satellite link; however, it minimizes the amount of data that would need to be retransmitted when packets are dropped. This operation dramatically slows down a satellite connection.

Scheduling capacity as well as the ‘‘burstiness’’ of subscriber traffic over the satellite link is another area that can potentially reduce data throughput or network efficiency. There are two common ways of scheduling capacity: connection-oriented and connectionless, which reflect two different philosophies of link management. Traffic burstiness determines which scheduling method is appropriate for data transmission.

The connection-oriented method assigns a fixed level of capacity prior to transmission and reassigns capacity at the completion of a transmission. What is remarkable about this method is that at least one round-trip delay is required to both allocate and de-allocate capacity. For long connections such as voice phone calls, the overhead to set up and tear down links is very small. For short connections, such as a URL request over a network, the setup and teardown time can overwhelm any efficiency the existing protocol achieves. The traditional switched telephone network’s traffic is relatively smooth and is well suited to this method. The connection-oriented method does not handle bursty traffic efficiently because bursts potentially slow down by the ceiling on capacity allocation. Connectionless refers in part to the way in which capacity is allocated. It requires no setup and teardown time. An underlying connectionless method is a multiple-access scheme that allows for random traffic, collision detection, and an efficient method for handling collisions to avoid channel breakdown. Traffic characteristic of data networking environments tends to arrive in bursts, which is suited to a connectionless method. The issue is how to run an efficient multiple-access scheme that addresses the issue of channel breakdown.