A tutorial on cross-layer optimization in wireless networks
read more
Citations
A tutorial on decomposition methods for network utility maximization
Layering as Optimization Decomposition: A Mathematical Theory of Network Architectures
Layering as optimization decomposition: A mathematical theory of network architectures : There are various ways that network functionalities can be allocated to different layers and to different network elements, some being more desirable than others. The intellectual goal of the research surveyed by this article is to provide a theoretical foundation for these architectural decisions in networking
Optimal energy management policies for energy harvesting sensor nodes
A distributed CSMA algorithm for throughput and utility maximization in wireless networks
References
Combinatorial optimization: algorithms and complexity
Rate control for communication networks: shadow prices, proportional fairness and stability
Stability properties of constrained queueing systems and scheduling policies for maximum throughput in multihop radio networks
Information capacity and power control in single-cell multiuser communications
Fair end-to-end window-based congestion control
Related Papers (5)
Stability properties of constrained queueing systems and scheduling policies for maximum throughput in multihop radio networks
Frequently Asked Questions (13)
Q2. What is the only assumption that the authors make on the interference relationship?
The only assumption that the authors make on the interference relationship is that it is symmetric, i.e., if link l interferes with link k, then l also interferes with k.
Q3. What is the way to solve the optimal user rate-vector?
The authors can then associate Lagrange multipliers for each constraint in (15), and solve the optimal user rate-vector ~x in Λ0 that maximizes the total system utility.
Q4. What is the common way to reduce the probability of collisions?
To reduce the probability of collisions, one may have to choose M sufficiently large which may again lead to a significant overhead.
Q5. What is the way to solve the problem of the node-exclusive interference model?
The so-called node-exclusive interference model (i.e., the data rate of each link is fixed at cij , and the only wireless constraint is that each node can only communicate with one other node at any time), where the optimal schedule corresponds to a Maximum-Weighted-Matching problem [28], [29], [42].
Q6. What is the main class of problems that the authors have described so far?
Another important class of problems deals with the development of opportunistic scheduling schemes with the intention of accommodating the maximum possible offered load on the system without violating stability or other QoS constraints.
Q7. What is the simplest way to say that a system is stable under a scheduling policy?
Given a user rate vector ~x = [xs, s = 1, ..., S], the authors say that a system is stable under a scheduling policy if the queue length at each node remains finite.
Q8. What is the way to generalize the node-centric formulation to the case?
As a final note, the node-centric formulation can also be generalized to the case with pre-determined routing, and the link-centric formulation can also be generalized to the case with multi-path routing [28].
Q9. What is the second approach to imposing the minimum guaranteed throughput?
A second approach is to impose the minimum guaranteed throughput as a resource constraint and use this constraint to generate congestion signals.
Q10. What is the way to show that the worst-case throughput loss is bounded?
In the previous subsection, the authors showed that, under the nodeexclusive interference model, the worst-case throughput loss is bounded by a factor 1/2 independent of the topology of the network.
Q11. What is the approach to solving the congestion problem?
The approach is to find a feasible rate region that has a simpler set of constraints similar to that of wireline networks and then develop congestion controllers that compute the rate allocation within this simpler rate region.
Q12. What is the problem with the scheduling component (8)?
The scheduling component (8) or (12) is usually difficult to solve because the rate power function u(·) in many wireless settings is not concave.
Q13. How can the authors use these algorithms to solve the congestion-control problem?
These scheduling algorithms can then be used to solve the joint congestion-control and scheduling problem, either via the layered approach in Section IV-A, or via the cross-layer approach in Section IV-B [29].