A high throughput path metric for multi hop wireless routing

This paper describes a new metric to be used for routing protocols to determine the best path to route in a wireless environment. Previously used metric is the minimum hop count, which favors the least hop path. This has several disadvantages. First, the paths now might be longer, because it takes the least amount of hops, this degrades signal quality since the longer the path, the more the signal is attenuated. This also maximizes loss ratio.

The new metric introduced in this paper is the expected transmission count. This metric takes into account the retransmission counts, and finds the effective transmission count. This allows the mtric to find the path with high throughput, even in the midst of losses. The ETX for a link is calculated by two different ratios, categorized as the forward and reverse ratio. The forward ratio is the ratio that data arrives at the recipient. The reverse ratio measure the ratio of the ack is received. By obtaining the ratio of these 2 values, the metric can then calculate the effective transmission count. The two values are obtained by dedicated link probe packets, which each node broadcasts at a fixed rate. Then depending on whether or not a node receives a probe packet at that rate, we can then calculate the loss ratio based on the last probe packet we received.

This metric takes into account link quality, along with the hop count. This is essential for wireless connections, where link layer losses are common. The minimum hop count path now may not be the ideal in this situation. Several questions arose in my mind when reading the paper. Based on the design of the protocol, it seems that sending the probe packets at that fixed rate is the key to determining the link quality. However, there seems to be ways to delay the sending of a probe packet, for example, if a large data packet is being transmitted etc. This issue will effectively skew the link quality and might lead to false information.