Should we change the rules for trip identification for GPS travel records?

GPS data processing usually consists of trip identification (TI), mode detection and purpose imputation. Given that most research needs to process millions of data points, it is always debatable what the optimal interval is to record data. In practice, most research uses one second as the interval, whilst some researchers use 3 seconds, 5 seconds, or an even longer interval to record data. Also, it is usually suggested that 120 seconds of dwell time would be a reasonable criterion to identify a stop. However, some activities, such as pickup/drop-off, may have a shorter duration. This paper investigates both issues in depth. This paper also discusses the trade-offs of choosing different options, suggests the relatively better option and the ways to cope with the trade-offs.