Conclusion

For any proposed extension to solve the wireless TCP problem, it is important that implementations be available to test changes to the algorithm. There are always hidden assumptions that simulators do not bring out, and complex interactions that occur in real systems. Any science with depends on simulation must continually benchmark theory and simulation against experimental evidence. Implementations also need to undergo the same kind of peer review that theory does, and must be aware of the latest theories and try to make use of them.