1: \begin{abstract}

2:   This invited paper presents some novel ideas on how to enhance the performance of

3:   consensus algorithms in distributed wireless sensor

4:   networks, when communication costs are considered. Of particular interest

5:   are consensus algorithms that exploit the broadcast property of the wireless

6:   channel to boost the performance in terms of convergence speeds. To this end, we

7:   propose a novel clustering based consensus algorithm that exploits

8:   interference for computation, while reducing the energy consumption in the

9:   network. The resulting optimization problem is a semidefinite program, which

10:   can be solved offline prior to system startup. 

11: %  are adaptive methods where consensus

12: %  algorithms are used to perform projections on some manifolds defined by a

13: %  target application. We show how the broadcast property of the wireless

14: %  channel can be exploited to boost the performance of consensus

15: %  algorithms. To this end, we propose a novel application-aware cluster-based

16: %  consensus algorithm that exploits interference for computation, while

17: %  reducing the number of measurements. When designing the algorithm,

18: %  communication costs will be taken into account. Some recent results are

19: %  applied to deduce the convergence of adaptive subgradient methods that

20: %  harness interference for computation.

21: \end{abstract}

22: