Audi-tory sensitive

Working in an open-plan office can be challenging. It is difficult to distinguish between your thoughts and the interesting, often heated discussions in the background .  Removing your headphones compounds the problem.

When you think of the highway as an open-plan office and allow cars to talk with each other, an important challenge is to coordinate communication so as to not have the available communication medium saturated with discussions on topics that may include which driver performs the best, or which turnpike has the biggest lanes, or how every car should read Fiat-effect and how it  is revolutionizing social networking.

To coordinate communication between network devices a protocol is required, part of which is a Medium Access Control (MAC) layer in the communication stack.  The MAC essentially determines who has the right to speak when, think of it as “the rules of engagement”.  The analogy in the class-room environment with the audible medium can be summarised as :  The teacher has the right to speak and a student can request the right to speak by raising their hand. The right to speak is given by the teacher looking at the student in question and nodding at them with an expectant facial expression. The student can then speak until finished or until interrupted by the teacher. The student indicates completion by pausing for a second or two and making eye contact with the teacher with an expectant facial expression. The school master can interrupt anyone with an announcement over the PA system. From this example, we see there are certain  ”protocol” to be followed, and there are priorities given to certain messages/messengers. This example is a fairly formal communication model, with strict rules. When people in a group are just casually chatting on the other hand, interruptions are more common and everyone can speak when they want to, within certain limitations of social acceptability.

In the vehicle scenario, the MAC could allow any node to just transmit whenever they feel like it and based on a response decide whether the message was successfully delivered, or whether it collided with another message from another car and was lost. This type of MAC is called collision-based. On the other hand, a MAC could be very rigid and structured, which allows only one node to speak at a specific time (called time division multiple access) or on a specific frequency (called frequency division multiple access) or combinations of these. These types of MACs are called collision-free. Collision-based MACs deliver better overall network performance for low network traffic loads but are susceptible to unbounded delays. Collision-free approaches work  better when the load is high,but wastes time slots if not fully occupied. In practice very few MACs are quite as simple and normally these fundamental approaches are improved upon to provide more intelligent MAC models. The 802.11 Wi-Fi model (de facto vehicular model), for example is based on the collision-based approach, but has build in “request to send” and “clear to send” messages, as well as a network allocation vector, that provide some collision avoidance. The more complex MACs mixes the two with time slices that are collision-based and time slices that are collision-free. Even though these approaches are theoretically more efficient, they are also much more complex and difficult to implement.

It should be obvious that communications between cars happen wireless. The implication is that there must be a certain threshold in RF power below which a receiver cannot distinguish between noise and the transmitted signal.  Neglecting obstacles for a second, this limitation reduces to a separation range, beyond which nodes cannot communicate.

Given the MAC and range challenged mentioned above, a new problem now arises: Who is in control of the protocol?In traditional Wi-Fi scenarios the access point (wireless router) in in control of synchronisation. In the case of vehicles the control could either be through a centralised entity (such as the infrastructure via road side units), self achieved through an externally received signal (e.g. GPS), or completely distributed such that only the vehicles in range of each other sort it out on their own.  In an urban environment road side units could be within range to control the MAC, but in rural areas or remote highways the vehicles are likely to be out of range of infrastructure and a distributed control would be required, albeit with a synchronisation nudge from the nigh-ubiquitous GPS.

And this is more or less what I am looking at. How do you get an unknown number of vehicles in a remote area to self-orchestrate communications in a way that is fair and flexible, without a lot of overhead, able to carry high levels of traffic. This MAC must be able to deal with the severe mobility of the nodes that go into and out of range of each other. It must also be able to support emergency messages.

No comments yet.

Leave a comment

Leave a Reply