Predicting Network Availability Using User Context

Abstract

Most mobile devices nowadays can simultaneously connect to different access networks with different characteristics at different times. Most solutions proposed for such an environment are reactive in nature. For example, when networks are encountered, the device performs a vertical handover to the network that offers the highest bandwidth. But the cost of handover may not be justified if that network is only available for a short time. Knowledge of future network availability and its capabilities would help to proactively handle the handover process more intelligently. Network availability prediction is often addressed as user path predictions with network coverage maps. In contrast, we model it as a more robust context prediction problem that can use any of the available context variables like GSM cell ID, WLAN AP, whether the power cable plugged, number of people around etc.Specifically, we propose a Semi-Markovian context prediction model to predict WLAN availability. As collecting and processing context consumes power, we propose a method to rank each context variable according to their contributions to prediction accuracy. We also employ the same method for optimizing model parameters. Real user data collected in our experiments show that when WLAN status is static, prediction errors are nearly zero and even in changing environments, error is less than 26% on average and the obtained context variable ranking is realistic. KeywordsNetwork Availability Prediction, Context Prediction, Semi Markov Model •We model availability prediction as a robust, context sensor agnostic prediction problem which uses any available context information like GSM cell IDs, Wi-Fi AP presence, whether LAN is connected, whether Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

Realistic data transfer scheduling with uncertainty

“…Our own prediction work [2,3] shows that even with a good prediction model, average prediction error is in the vicinity of 20%, which is significant. For example, we may predict that Wi-Fi will be available 60% of the time, but it might as well be 40% available or 80% in reality, which are called different availability scenarios.…”

“…We observed in [2,3] that the predictions cannot be done for arbitrarily longer periods with good accuracy. Predictions of network availability can be achieved within acceptable accuracy limits when the durations concerned are short (i.e.…”

“…We addressed that in our previous work and proposed several prediction models [2,3]. With that experience and from others work [29], we make two key observations which need to be addressed in the scheduling, and are described in the first two sections below.…”

“…We predicted the Wi-Fi availability for a 5 min duration using all the previous days' data plus data of the same day just before the prediction time with a learning process, and we did this repeatedly for all the 5 min blocks in a day, for the last 5 days of each user. Interested readers are referred to [2,3] for more details.…”

“…Fifteen days worth of data which included Wi-Fi presence and GSM presence recorded for every half a minute were gathered from each volunteer [2,3]. GSM was almost always available for all the users as opposed to Wi-Fi.…”