Topology Management

Weihang Jiang Weili lin Vishvesh Sahasrabudhe

Outline

The function of “topology management” Different approaches for topology managements Several criteria of a “good” topology

management “Problems” will be addressed now and then

The function of topology

Scheduling of transmission, routing, broadcasting, spatial querying.

Main issue is power saving, and maintaining connected (is “connected” necessary?)

Different Approaches

Topology Management

Power control Hierarchical

One - hop d-hops

Deterministic

(TMPO)

Non-deterministic

(SPAN, Self-organization)

K-connectivity 1-connectivity

Several criteria of a “good” topology management The most important criterion is power saving

Fewer cluster heads Load balance Average degree (power control)

Other criterion Connectivity (SPAN, K-connectivity) Algorithm convergence speed (self-organization) Stability (fewer nodes changing their role: delay

changes, how to decide the cycle?) Mobility (always think about how to deal with the impact

of movement, can we take advantage of it?) Fault tolerant

Fewer cluster heads

Method 1, select a lot of cluster heads to let all cluster heads connected, then withdraw unnecessary cluster heads to hosts

Method 2, select minimum dominating set, then add gateways/doorways to connect them

Method 3, using greedy algorithm, then on each phase, the cluster heads are connected

Method 4 ???

Load balance

Method 1, Force long-term working cluster heads back to host nodes and sleep

Method 2, Select cluster heads basing on battery level (maybe partially)

Power Control

Adjusting transmission range of nodes to achieve both connectivity and reduction of energy consumption

1-connectivity

RA assignment problem

1. RA: determine RA such that the total cost of power is minimum

2. HRA:determine the minimum RA such that r-homogeneous assignment is connecting

3. KRA:determine the minimum RA such that is connected

Problem?

Directional links

A B AB

Distance estimation

RSSi: radio signal strength(poor accuracy)

Toa: time of arrival

Preferred K

Mobility

Rely on volatile information(e.g. distance)

K-connectivity

Motivation: fault tolerant

Theoretical background

Main problems

Interesting problems

Relationship between k-connectivity and k minimum degree

Power Management using Span

Requirements of power-saving coordination algorithm Max idle nodes Minimal delay in packet forwarding Non-interference of original paths No assumption of link-layer’s facilities for

sleeping Interoperate correctly with routing system

Span Algorithm

A set of coordinators on for topology maintenance

Coordinator election based on connectivity and node power

Span Algorithm

Local Coordinator election Based on Coordinator eligibility Two neighbors cannot reach each other – node

becomes a coordinator Announcement Contention resolved by back-off

delay

Span Algorithm

Delay Calculation based on connectivity and available power

Leads to efficient management of power and approx. maximum connectivity with minimum nodes