Sichao Yang

Optimal Delay–Throughput Tradeoffs in Mobile Ad Hoc Networks

In this paper, we investigate the delay-throughput tradeoffs in mobile ad-hoc networks. We consider four node mobility models: 1) two-dimensional independent and identically distributed (i.i.d.) mobility, 2) two-dimensional hybrid random walk, 3) one-dimensional i.i.d. mobility, and 4) one-dimensional hybrid random walk. Two mobility time scales are included in this paper. i) Fast mobility, where node mobility is at the same time scale as data transmissions. ii) Slow mobility, where node mobility is assumed to occur at a much slower time scale than data transmissions. Given a delay constraint D , we first characterize the maximum throughput per source-destination (S-D) pair for each of the four mobility models with fast or slow mobiles. We then develop joint coding-scheduling algorithms to achieve the optimal delay-throughput tradeoffs.

Paging and Registration in Cellular Networks: Jointly Optimal Policies and an Iterative Algorithm

This paper explores optimization of paging and registration policies in cellular networks. Motion is modeled as a discrete-time Markov process, and minimization of the discounted, infinite-horizon average cost is addressed. The structure of jointly optimal paging and registration policies is investigated through the use of dynamic programming for partially observed Markov processes. It is shown that there exist policies with a certain simple form that are jointly optimal. An iterative algorithm for policies with the simple form is proposed and investigated. The algorithm alternates between paging policy optimization, and registration policy optimization. It finds a pair of individually optimal policies. Majorization theory and Rieszs rearrangement inequality are used to show that jointly optimal paging and registration policies are given for symmetric or Gaussian random walk models by the nearest-location-first paging policy and distance threshold registration policies.

VCG-Kelly Mechanisms for Allocation of Divisible Goods: Adapting VCG Mechanisms to One-Dimensional Signals

The well known Vickrey-Clark-Groves (VCG) mechanism provides socially optimal solutions for many allocation problems with strategic buyers, but for divisible goods the bids are infinite dimensional. F.P. Kelly and his co-workers developed an allocation mechanism based on one dimensional bids, which is socially optimal if the buyers are price-takers. The idea is that the one-dimensional bid from a buyer specifies a surrogate valuation function. We propose the VCG-Kelly mechanism, which is obtained by composing the one-dimensional signaling idea of Kelly with the VCG mechanism, providing socially optimal allocation for strategic buyers at the Nash equilibrium point. The VCG-Kelly mechanism is studied in the case of a network rate allocation problem, and it applies to several others. It is shown how the revenue to the seller can be maximized or minimized using a particular one-dimensional family of functions. The Nash equilibrium point of the mechanism is shown to be globally stable.

Paging and registration in cellular networks: jointly optimal policies and an iterative algorithm

This paper explores optimization of paging and registration policies in cellular networks. Motion is modeled as a discrete-time Markov process, and minimization of the discounted, infinite-horizon average cost is addressed. The structure of jointly optimal paging and registration policies is investigated through the use of dynamic programming for partially observed processes. It is shown that there exist policies with a certain simple structure that are jointly optimal, though the dynamic programming approach does not directly provide an efficient method to find the policies. An iterative algorithm for policies with the simple form is proposed and investigated. The algorithm alternates between paging policy optimization and registration policy optimization. It finds a pair of individually optimal policies, but an example is given showing that the policies need not be jointly optimal.

A mechanism for pricing service guarantees

The calculus of deterministic constraints on service and traffic streams offers a rich language to specify service guarantees. In particular, the service curve earliest deadline first (SCED) algorithm has an associated feasibility test given by linear constraints. Users sending streams of data through the server may have differing needs for delay and throughput.We suggest a way based on utility function maximization, subject to the linear constraints of the SCED algorithm, for allocation of service. In addition, a generalization of the SCED algorithm is given which does not require that the deadline sequences within streams be monotone nondecreasing.