The Internet, and to a degree wireless networking developments, are an attractive source of inspiration for NASA engineers. After all, with budget constraints on everyones minds, if lessons can be learned from existing technologies that prevent potentially expensive mistakes, so much the better. Starting from scratch is always an expensive option, but can Internet protocols and strategies really be launched into space? A cursory look at the practicalities would seem to suggest not, with Mars being some 34.6 million miles away from Earth, at least. The actual distance is changing all the time, and even at the speed of light radio signals will take at least three minutes to arrive, and perhaps as long as 20 minutes. Realistically, the round-trip times are likely to be much longer, allowing for the relative positioning of Earth and Mars around the Sun, and could take anything between 15 and 50 minutes. Latency on this scale just cant work with current Internet protocols, which need to acknowledge safe receipt of data packets or request re-transmission if unsuccessful - such acknowledgements being measured in milliseconds rather than minutes. So with a space speed limit firmly in place, where data is looking at round-trip times in excess of ten minutes, a new set of protocols will need to be developed to link together a cosmic network of Internets.
This isnt as simple a task as it may sound, even in these days of technological advance and network wizardry. Handling the excessive round-trip delays is perhaps the biggest problem, but others such as overcoming the data stream corruption or obscurity when crossing the interplanetary divide caused by background noise are just as tricky. Even the DSN antenna arrays suffer from transmission errors caused by this cosmic interference, and are dealt with by the successful if somewhat brutal method of sending everything twice. The thinking being that by sending everything in duplicate, the odds are stacked high in favour of the same bits of the broadcast not being lost both times. A terrestrial computer then compares both data streams and pieces them together to complete the transmission picture.
There are, it would appear, two schools of thought as to the best way forward in developing interplanetary Internet-working protocols. One approach is to use a bundle space model, where instead of using a packet transport system as were used to with our Earthbound Internet, a single data transaction methodology is implemented. This would involve all the broadcast information being bundled together with the necessary data to process it at the other end. The idea being that each planet that joins the IPN would have a gateway to handle data traffic, meaning an information bundle could arrive at any such gateway where it would get routed to the next, and onwards slowly to its destination. The DSN could become the default Earth gateway in such a scenario, for example.
But while the DSN may survive in this model, its likely that the DNS (Domain Name System) wouldnt - at least, not as we know it today. With severe restrictions on bandwidth and memory in space, as already touched upon, it wouldnt be feasible to keep an active and up-to-date name server on each spacecraft. Instead, its much more likely that it would be up to each gateway to handle the delivery to the final address, with data just being routed to that gateway in the form of earth.sol or mars.sol domains. Indeed, I understand that NASA has already, through JPL, approached the ISO (International Standardisation Organisation) in Geneva to tentatively enquire as to the realities of setting up a solarsystem .sol superdomain for earth.sol and moon.sol. It has
been reported that the response boiled down to the requirement of needing a sovereign representative of each planet involved to provide written letters of intent! The bundle-space approach isnt without its own inherent problems,
like getting around the fact that celestial objects move constantly in relation to each other, so the communication paths will be rather complex, to say the least - unlike security, which couldnt be complex as bundle space isnt an interactive transport medium, so no public key exchange crypto is necessary. Instead, a combination of pre-shared keys and limited access to the network would be the most likely solutions.
The latter is almost certainly going to be the case in any given scenario for a while, as limited bandwidth, cost of implementation and the value of data transmitted all add up to the simple fact that spamming email@example.com isnt likely to be an option. It seems to me that ultimately the alternative interplanetary Internet-working protocol methodology, which is being developed by the JPL engineers responsible for standardisation of interplanetary NASA missions, will be more likely to succeed.
This would involve a packet-based protocol that actively monitors the delivery of data packets and requests missing ones for retransmission, but crucially, and unlike our existing TCP/IP, it also has the ability to keep on running despite an incomplete stream of packets. This protocol would allow for missing packets to be transmitted days later, with no effect on the eventual data assembly and packet processing procedures. This model also does away with the complex address headers that we see in TCP/IP, which consume 500 bits a time, far too big in space data-transmission terms. Instead, a stripped-down header element is being developed that totals a mere two per cent of each data broadcast, which equates to around 48 bits for each header.
Unfortunately, this would mean that the new protocol wouldnt be compatible with the Internet, but thats a small price to pay for the extra data that could be transmitted in each packet. And anyway, Internet compatibility will have to take a backseat in the early days of any Mars Network, perhaps coming into play when Vinton Cerfs IPN starts taking shape some 30 or 40 years down the road. Having said that, JPL foresees a system whereby an additional terrestrial gateway could act as an interpreter between the protocols, allowing NASA scientists to access the network by way of its Internet connections.
This Feature appeared in the June, 2000 issue of PC & Tech Authority Magazine