Location-based services (LBSs) via mobile handheld devices have been subject to major privacy concerns for users. Currently, most of the existing works concerning the continuous LBS queries mainly focus on users' privacy demands with little consideration of the service of quality (QoS). In this paper, we propose a demand-aware location protection scheme for continuous LBS requests, allowing a user to customize not only location privacy but also QoS requirement, while this results that in considerably many queries points, the privacy and QoS requirement cannot be met together, and the location privacy protection cannot be provided for the continuous LBS queries. We point out that its underlying reason is that in few LBS query regions, the footprints are sparse or the privacy requirements are set unreasonably high. Therefore, a maximum demands-aware query sequence algorithm is proposed in the scheme. Through identifying and restraining the queries in those regions, most of LBS queries are satisfied; thus, the longest LBS query sequence is obtained, which can satisfy a user's specific privacy and QoS requirements simultaneously. Furthermore, on the premise that the user's privacy requirement is met, in demand-aware location protection scheme, we propose two algorithms to minimize the constructed cloaking regions, reducing the query latency and the server's workload and providing better QoS for users. Extensive simulations on a large dataset prove the effectiveness of our approach under various location privacy and QoS demands. DALP: A DEMAND-AWARE LOCATION PRIVACY PROTECTION SCHEME 4.2.1. Finding the queries where the footprints are sparse in the QoS-constrained regions. In the continuous LBS queries Q D ¹Q 1 ; Q 2 ;; Q n º, if there is a query point Q i where the distribution of footprints in its QoS-constrained region A i is sparse, it may lead the number of common users between A i and A A i to decrease. And then, the privacy level that the continuous LBS queries can Figure 3. An example of continuous location-based service queries.provide will decrease, resulting in that the privacy demands at most query points cannot be satisfied. As shown in Figure 3, there are just two users' footprints at query point Q 1 , which will greatly limit the privacy levels of other queries under k-anonymity model.To find and constrain those queries, we give the footprint-sparse query region search algorithm. Based on the foregoing analysis, we know that constraining such kind of queries may significantly increase the number of common users in A A i , and thus the privacy levels of the rest queries will be increased saliently. In Algorithm 1, the query in the original LBS queries sequence Q will be constrained one by one, and the summed privacy level of other queries is calculated. During this process, we select a query Q s as the constraint point if the summed privacy of the rest queries can reach the maximum value after constraining it. At the same time, the query set S 1 that still cannot meet the privacy demands after the constr...