Wei-jen Hsu

Modeling Time-Variant User Mobility in Wireless Mobile Networks

Realistic mobility models are important to understand the performance of routing protocols in wireless ad hoc networks, especially when mobility-assisted routing schemes are employed, which is the case, for example, in delay-tolerant networks (DTNs). In mobility-assisted routing, messages are stored in mobile nodes and carried across the network with nodal mobility. Hence, the delay involved in message delivery is tightly coupled with the properties of nodal mobility. Currently, commonly used mobility models are simplistic random i.i.d. model that do not reflect realistic mobility characteristics. In this paper we propose a novel time-variant community mobility model. In this model, we define communities that are visited often by the nodes to capture skewed location visiting preferences, and use time periods with different mobility parameters to create periodical re-appearance of nodes at the same location. We have clearly observed these two properties based on analysis of empirical WLAN traces. In addition to the proposal of a realistic mobility model, we derive analytical expressions to highlight the impact on the hitting time and meeting times if these mobility characteristics are incorporated. These quantities in turn determine the packet delivery delay in mobility-assisted routing settings. Simulation studies show our expressions have error always under 20%, and in 80% of studied cases under 10%.

Modeling spatial and temporal dependencies of user mobility in wireless mobile networks

Realistic mobility models are fundamental to evaluate the performance of protocols in mobile ad hoc networks. Unfortunately, there are no mobility models that capture the non-homogeneous behaviors in both space and time commonly found in reality, while at the same time being easy to use and analyze. Motivated by this, we propose a time-variant community mobility model, referred to as the TVC model, which realistically captures spatial and temporal correlations. We devise the communities that lead to skewed location visiting preferences, and time periods that allow us to model time dependent behaviors and periodic reappearances of nodes at specific locations. To demonstrate the power and flexibility of the TVC model, we use it to generate synthetic traces that match the characteristics of a number of qualitatively different mobility traces, including wireless LAN traces, vehicular mobility traces, and human encounter traces. More importantly, we show that, despite the high level of realism achieved, our TVC model is still theoretically tractable. To establish this, we derive a number of important quantities related to protocol performance, such as the average node degree, the hitting time, and the meeting time, and provide examples of how to utilize this theory to guide design decisions in routing protocols.

Weighted waypoint mobility model and its impact on ad hoc networks

To realistically evaluate performance of ad hoc networks we propose a generic framework called the Weighted Way Point (WWP) mobility model. WWP model captures preferences in choosing destinations of pedestrian mobility patterns in a campus environment. We estimate the parameters of this model using mobility survey data for the USC campus. We further compare WWP model with widely used Random Waypoint (RWP) model and demonstrate that in the WWP model mobile nodes display uneven (clustering), time-varying spatial distribution. WWP model is also less mobile than RWP model with typical parameter settings. The clustering effect can cause lower success rate of route discovery in ad hoc networks.

Profile-cast: behavior-aware mobile networking

In this paper we advocate a service paradigm, profile-cast, within the communication framework of delay tolerant networks (DTN) (K. Fall, 2003). This novel approach leverages the behavioral patterns of mobile network users for delivering messages to a sub-group of users as defined by their profiles (e.g., interest, social affiliation, etc.). We study large data sets of user mobility profiles and present a case-study of mobility profile-cast with a similarity-based forwarding protocol. We show that behavior-aware protocol design has a great potential - we reduce the total number of transmissions to 45% of flooding under 92% delivery success rate, or to only 3% transmissions of flooding under 61% delivery success rate. It also leads to shorter delay (at least 30% less) as compared to a random transmission protocol.

On Nodal Encounter Patterns in Wireless LAN Traces

In this paper, we analyze multiple wireless LAN (WLAN) traces from university and corporate campuses. In particular, we consider important events between mobile nodes in wireless networks-encounters. We seek to understand encounter patterns in the mobile network from a holistic view, using a graph analysis approach. Such an analysis sheds light on the diverse, nonhomogeneous nature of users in the given environments in terms of their encounter events with other nodes. Furthermore, we evaluate the feasibility of forming an infrastructureless network to reach most of the nodes utilizing time-varying internode connectivity through encounters, and the robustness of such an ad hoc communication network. Our analysis shows that while the encounter events are “sparse” (i.e., any given node does not encounter with many other nodes), the connectivity of the whole network is well-maintained, and a Small World pattern of nodal encounter emerges for the observation periods longer than one day. More interestingly, the encounter events collectively form a robust communication network, in which store-carry-forward message dissemination can be successful even with over 20 percent noncooperative nodes or removal of short-lived (up to minutes) encounter events.

On Modeling User Associations in Wireless LAN Traces on University Campuses

In this paper we analyze wireless LAN (WLAN) traces collected from four different sources, including three university campus WLANs and one corporate WLAN to compare the similarities and differences of user association behavior to access points (APs) in these environments. This study provides extensive comparison of multiple WLAN traces, and outlines a basis for creating models for user association patterns in WLANs. We propose a set of important metrics for modeling the association patterns of wireless LAN (WLAN) users. Specifically, we look into (a) Activeness of users, (b) Macro-level mobility, (c) Micro-level mobility and (d) Repetitive association patterns. We find that (1) A significant portion of users are offline for non-neligible fraction of time (on average, the online time fraction is between 87.68% and 14.12% for the traces we studied). (2) Users visit only a small subset of APs (on average less than 5%, and the maximum is less than 35%), and (3) Users show periodic pattern of visiting the same APs in some traces. The findings along these aspects show similar trends among the traces, with differences in details due to both underlying user population/environments and methodologies of trace collection.

A local metric for geographic routing with power control in wireless networks

We investigate the combination of distributed ge- ographic routing with transmission power control for energy efficient delivery of information in multihop wireless networks. Using realistic models for wireless channel fading as well as radio modulation and encoding, we first show that the optimal power control strategy over a given link should set the transmission power to achieve a special signal-to-noise ratio (SNR) constant that can be computed using an elegant characteristic equation. Counter-intuitively, for typical radios, this corresponds to an optimal operating point of SNR that lies in the transitional region (where packet error rates are non-negligible). We then propose a local power efficiency metric for distributed routing such that at each step the transmitter picks as the next hop the neighbor for which this metric is maximized. Through extensive simulations, we compare the performance of the proposed algorithm and globally optimal routing algorithms. We show that in randomly deployed 2-D networks, the combination of this local metric for routing with optimal power control has close performances, in terms of average power consumption under different node density settings and physical transmission power limits, to the best strategy using global network link state information. In particular, when electronic power is relatively low, the proposed algorithm can provide up to six times reduction in power usage compared to channel-unaware routing algorithms.

CSI: A paradigm for behavior-oriented profile-cast services in mobile networks

We propose a new behavior-oriented communication paradigm in mobile networks, profile-cast, motivated by tight user-network coupling in mobile societies. In this novel paradigm, messages are sent to sender-specified target profiles, instead of machine IDs. We present a systematic framework for such services. First, we analyze the spatio-temporal stability of user mobility profiles constructed from empirical data sets, and they turn out to be surprisingly stable. The similarity of the current mobility profile of a user to its future mobility profile remains above 0.6 for five weeks, while the correlation coefficient of the similarity metrics between a user pair at different time instants is above 0.5 for two weeks. Second, we present a protocol for the profile-cast service, named CSI, and provide a fully distributed solution utilizing behavioral profile space gradients and small world structures to selectively diffuse information across the network towards the intended recipients. Leveraging stability in user behaviors, the two modes of CSI achieve good performance compared to the theoretical optimal protocols. Both CSI:Target mode and CSI:Dissemination mode achieve more than 94% delivery ratio. Comparing with the delay-optimal protocol, they show no more than 47% and 32% more delay, respectively, with at most 10% more transmission overhead. Comparing with the overhead-optimal protocol, they use no more than 7% more overhead while achieving dramatic improvement in delay (up to 150% less). Both CSI:T and CSI:D significantly outperform the epidemic routing, using less than 7% overhead, and variants of random walk, where CSI:T doubles the delivery ratio using less overhead, and CSI:D shows at least 50% less delay under similar overhead. We believe the profile-cast paradigm would enable many behavior-oriented services efficiently, such as targeted announcements and profile-based alert notifications, in various mobile networks.

Profile-Cast: Behavior-Aware Mobile Networking

In this paper we advocate a service paradigm, profile-cast, within the communication framework of delay tolerant networks (DTN) (K. Fall, 2003). This novel approach leverages the behavioral patterns of mobile network users for delivering messages to a sub-group of users as defined by their profiles (e.g., interest, social affiliation, etc.). We study large data sets of user mobility profiles and present a case-study of mobility profile-cast with a similarity-based forwarding protocol. We show that behavior-aware protocol design has a great potential - we reduce the total number of transmissions to 45% of flooding under 92% delivery success rate, or to only 3% transmissions of flooding under 61% delivery success rate. It also leads to shorter delay (at least 30% less) as compared to a random transmission protocol.

On the efficacy of mobility modeling for DTN evaluation: Analysis of encounter statistics and spatio-temporal preferences

In mobile networking, the main goal of mobility modeling and simulation is the ability to accurately reproduce effects of realistic mobility on the performance of networking protocols. In the areas of adhoc and delay tolerant networks (DTNs), recent work on mobility modeling focused on replicating metrics of encounter statistics and spatio-temporal preferences. No studies have been conducted, however, to show whether matching these metrics is sufficient to accurately reproduce DTN protocol performance. In this study, we address this specific problem, and attempt to show the sufficiency (or lack thereof) of existing encounter and mobility metrics in reproducing realistic effects of mobility on networking protocols. We first analyze the characteristics of two well-established mobility models; the random direction and the time-variant community (TVC) models, and study whether they capture encounter statistics and preference patterns observed in real-world traces. Second, we contrast the performance of epidemic routing in DTNs based on the mobility models, to that based on extensive mobility traces. We provide two main findings. First, careful parameterization of the models can indeed replicate the metrics in question (e.g., inter-encounter time distribution). Second, even carefully crafted mobility models surprisingly result in protocol performance that is dramatically different from the trace-driven performance. The difference in message delivery delays can reach 67%, while difference in reachability approaches 80%. Such findings strongly suggest the need to revisit mobility modeling. Furthermore, they clearly show the insufficiency of existing encounter and preference metrics as a measure of mobility model goodness. Systematically establishing a new set of meaningful mobility metrics should certainly be addressed in future works.