Evangelia Kokolaki

Leveraging Information in Parking Assistance Systems

This paper systematically explores the efficiency of uncoordinated parking space allocation in urban environments with two types of parking facilities. Drivers decide whether to go for inexpensive but limited on-street parking spots or for expensive yet overdimensioned parking lots, incurring an additional cruising cost when they decide for on-street parking spots but fail to actually acquire one. Their decisions are made under perfect knowledge of the total parking supply and costs and different levels of information about the parking demand, i.e., complete/probabilistic information and uncertainty. We take a game-theoretic approach and analyze the parking-space allocation process in each case as resource selection game instances. We derive their equilibria, compute the related price-of-anarchy (PoA) values, and study the impact of pricing on them. It is shown that, under typical pricing policies on the two types of parking facilities, drivers tend to overcompete for the on-street parking space, giving rise to redundant cruising cost. However, this inefficiency can be alleviated through systematic manipulation of the information that is announced to the drivers. In particular, counterintuitive less-is-more effects emerge regarding the way that information availability modulates the resulting efficiency of the process, which underpin general competitive service provision settings.

Value of information exposed: Wireless networking solutions to the parking search problem

Our work draws on a concrete parking space search application to explore fundamental tradeoffs of wireless networking solutions to the provision of real-life services. In particular, we consider a city area, wherein each vehicle (mobile user) moves towards a chosen destination and seeks vacant parking space in its vicinity. Three main approaches to the parking space search problem are investigated, each representing a distinct paradigm of how wireless networking communications can assist the information management process. In the first approach, the vehicles execute the currently common “blind” sequential search for parking space by wandering around the destination. In the second distributed approach, the vehicles, while moving around the area, opportunistically collect and share with each other information on the location and status of each parking spot they encounter. Finally, with the third approach, the allocation of parking spots is managed by a central server availing global knowledge about the parking space availability. We compare the three approaches with respect to the time and distance vehicles need to travel before they park as well as the proximity of the assigned parking spots to the travel destinations. Results obtained under two scenarios for the user travel preferences (uniformly distributed travel destinations vs. a single hotspot road), reveal that the relative performance of the three solutions can vary significantly and not always inline with intuition. In the hotspot scenario, the centralized system consistently yields the minimum times and distances at the expense of more distant parking spot assignments; whereas, when user travel destinations are uniformly distributed, the relative performance of all three schemes changes as the vehicle volume grows, with the centralized approach gradually becoming the worst solution. We discuss the way each approach modulates the information dissemination process in space and time and resolves the emerging competition for the parking resources. We also outline models for getting analytical insights to the behavior of the centralized approach.

Cognitive Structure of Collective Awareness Platforms

Collective awareness platforms (CAPs) are internet and mobile tools for collaboration, sustainability and social innovation that can allows drastic improvement of our lifestyle, beyond the standard economic model. However, their development is often driven (and motivated) by technology, while their adoption and usage characteristics are determined by the social interactions and can be affected by many items, up to failure. We describe here our approach to CAPs modelling that includes elements from cognitive and evolutionary sciences, in the hope of providing instruments for the improvement and the assessment of CAPs.

Vulnerability of opportunistic parking assistance systems to vehicular node selfishness

Abstract Opportunistic networking leverages the volume, heterogeneity and mobility of end user nodes to foster the dissemination of information in the absence of network infrastructure. Nevertheless, in competitive settings (where the possession of information itself is an asset) user nodes often face a strategic dilemma: cooperate, to realize the network and support the information flow, or not do so, to gain competitive advantage over the other nodes. In this paper, we investigate realistic scenarios of opportunistic parking assistance service that instantiate such dilemmas. Ideally the vehicular nodes opportunistically collect and share information on the location and availability status of the parking spots. Yet the competition for parking spots may give rise to various facets of misbehaviors, such as deferring from sharing information (free riders) and/or deliberately falsifying disseminated information (selfish liars) so as to divert others away from a particular area of own interest. Simulation results indicate that misbehaviors tend to reduce the distance between the destination and the occupied parking spot for all vehicles at the expense of higher parking search times. However, misbehaving nodes fail to obtain any substantial performance advantage that would indeed encourage their misbehaviors. The addition of Mobile Storage Nodes compensates for the reduction of the information flow due to free riders but has almost no effect against selfish liars. Simple analytical models drawing on mean-field arguments provide further evidence for the fundamental dynamics that emerge from the interaction of the vehicular nodes.

Parking Assisting Applications: Effectiveness and Side-Issues in Managing Public Goods

With the emergence of mobile communication devices and social networking applications, new opportunities arise for various mobile networking applications. In this paper, we seek to experimentally study some fundamental properties of vehicular social applications that have been deployed to assist in the parking search process. The awareness and incentive mechanisms that are commonly incorporated in different instances of social parking applications are modeled and simulation scenarios are considered to explore particular aspects of these applications. It is shown that application users experience improved performance due to the increased efficiency they generate in the parking search process, without (substantially) degrading the performance of non-users. This is extremely important since applications managing common (public) goods should not provide benefits to their users by penalizing or almost excluding non-users. The incentive mechanisms are effective in the sense that they do provide preferential treatment to those fully cooperating but they induce rich-club phenomena and difficulties to newcomers. Interestingly, those problems, that may be a concern for all applications managing common (public) goods, seem to be alleviated by free-riding phenomena and dynamic behaviors.

On the Efficiency of Information-Assisted Search for Parking Space: A Game-Theoretic Approach

This paper seeks to systematically explore the efficiency of the uncoordinated information-assisted parking search in urban environments with two types of parking resource facilities: inexpensive but limited facilities (public) and expensive yet unlimited ones (private); an additional cruising cost is incurred when deciding for a public facility but failing to actually utilize one. Drivers decide whether to go for the public or directly for the private facilities, assuming perfect knowledge of prices and costs, total parking capacities and demand; the latter information can be broadcast by an ideal centralized information dissemination mechanism, assisting the otherwise uncoordinated parking search process. Drivers are viewed as strategic decision-makers that aim at minimizing the cost of the acquired parking spot. We formulate the resulting game as an instance of resource selection games and derive its Nash equilibria and their dependence on the environmental parameters such as the parking demand and supply as well as the pricing policy. The cost at the equilibrium states is compared to that under the optimal resource assignment (dictated to the drivers directly by an ideal centralized scheme) and conditions are derived for minimizing the related price of anarchy. Finally, the numerical results and the presented discussion provide hints for the practical management and pricing of public and private parking resources.