The advent of embedded sensors and their low cost integration in handheld devices (e.g. smartphones) are making them increasingly aware of the human location and context. There have been attempts to extract certain gait features (e.g. step length, step frequency etc.) based on data recorded from handheld devices. However, these attempts have been mostly inspired from observations in biomechanics. Hence, there is a profound need to study the modeling of human walking gait cycle while taking into account the different device carrying modes. It is hypothesized that the presence of handheld device in one hand can alter the step level symmetry of human walking gait cycle without affecting the stride level symmetry. The aim of this paper is to present a model of human walking gait cycle in different device carrying modes over a stride, which is based on parametric optimization technique used in robotics motion generation and the results of a preliminary experimentation conducted using motion capture technology. Both simulation and pilot experiments confirm that the presence of a small mass in one hand can affect the step level symmetry of the human walking gait which constitutes the novel outcome of this paper. Overall, the model successfully captures human walking features and can stand useful for the enhancement of existing pedestrian navigation algorithms with handheld devices for an increased autonomy of elderly people and pedestrian's mobility in general.