Professional Documents
Culture Documents
Sensor Management
Viswanath Avasarala , Member, IEEE , Tracy Mullen , Member, IEEE,
David Hall, Senior Member IEEE, and Sudheer Tumu
Abstract—Distributed Sensor management, the process of target (task Ta). B is interested in searching for the
managing or coordinating the use of sensing resources in a presence of new targets in a certain region (task Tb). The
distributed environment, is a multi-objective optimization sensor manager should allocate its resources including
problem. In our earlier work, we proposed MASM (Market- sensor schedules and communication bandwidth to
Architecture for Sensor Management), a market-based optimize performance on both Ta and Tb. Approaches like
approach to allocate sensor resources in real-time to various dynamic goal lattices [5], decision theory [6], Bayesian
resource requestors. MASM models the multi-objective
networks [7] have had use in finding the priorities/weights
sensor management problem as a combinatorial-auction
based market where the network resources sell goods to the of various objectives in the sensor management
resource requestors. To allow the resource requestors to optimization. However, a comprehensive paradigm to
participate in the market, MASM grants “budgets” to these take mission objectives and the current environments state
resource requestors based on their priority to the overall simultaneously and directly into consideration during
mission. However, for a given budget, self-interested resource allocation is not available.
resource requestors or buyers can learn from market-data In our earlier work, we have proposed a market-based
and adapt their bidding behavior. This paper presents resource allocation method for sensor management called
results of an initial experimental study, where the learning MASM (Market-Architecture for Sensor Management)
behavior of resource requestors is modeled and their effect
[8]. MASM is based on the Market-Oriented
on market performance is examined.
programming technique, which uses market algorithms
for resource allocation in distributed environments [9].
MASM models the multi-objective sensor management
I. INTRODUCTION problem as a combinatorial-auction based market where
the network resources sell goods to the resource
Sensor management which may be defined as “a process, requestors. To allow the resource requestors to participate
which seeks to manage or coordinate the use of sensing in the market, MASM grants “budgets” to these resource
resources in a manner that improves the process of data requestors based on their priority to the overall mission.
fusion and ultimately that of perception, However, for a given budget, self-interested resource
synergistically”[1]. The recent advances in sensor and requestors or buyers can learn from market-data and adapt
communication technologies have led to a proliferation of their bidding behavior.
sensor network applications [2][3]. For example, The adaptive nature of the agent behavior due to their
distributed sensor networks that cater to multiple users learning from market-history has a direct impact on
which query the sensor manager using web-based market-based resource allocation. In this paper, we use a
communication are now feasible. Examples for this multi- simple representation of agent learning to model their
sensor, multi-user sensor management scenario include: adaptive behavior and study the impact on market-based
(1) network-centric warfare, in which multiple sensing resource allocation. The paper is organized as follows.
platforms, sensor nets, and individual soldiers with Section II introduces MASM architecture and explains
sensors interact to allow rapid tactical situation briefly the combinatorial-based mechanism used for
assessment and threat assessment [3, 4], and (2) Condition sensor management. Section III explains our approach to
monitoring of the environment via ground-based, airborne model agent learning based on market data. Section IV
and space-based sensing systems for assessing different shows the results and provides some concluding remarks.
issues like weather, pollution etc. In these scenarios, the
various users could be trying to accomplish different tasks II. MASM OVERVIEW
using network resources. Distributed sensor management
is a multi-objective optimization problem. For example,
assume that two resource requestors A and B are trying to The design of MASM is shown in Figure 1. The
use a sensor network to accomplish different two tasks. A Mission Manager (MM) is responsible for allocating task
is interested in estimating the position and velocity of a responsibilities and budgets to the various agents in the
market. The actual details of the sensor network are
invisible to the consumer agents. Instead, consumers are
allowed to bid on high-level tasks, like “Track Target X CCA uses a bid translation function to create low-level
within an accuracy limit Y”. The sensor manager (SM) resource bids from high-level task bids. For each time slot
stipulates the type of tasks that the consumers can bid on, t, the auctioneer constructs bids on each resource set, S,
so that the consumer bids are restricted to tasks SM knows that can be allotted to task X. To construct these bids, the
how to accomplish. Consumer bids are of the type <t, p> auctioneer needs to calculate the price associated with the
where t is the task description that includes the task type bid on resource set S, for task X, based on the consumer
and final task quality desired by the consumer, and p is bid price P. For this purpose, a novel mechanism of price
the price that the consumer is willing to pay. For calculation has been devised. For the task X, the
example, the task description for a bid to search a auctioneer computes the bid price for a resource set S as
particular grid, x, so that the uncertainty of target presence the percentage of the consumer task completed by the
(as measured by entropy) < 0.001 is as follows: resource set multiplied by the actual consumer bid price P.
(type: search Computation of the percentage of task completed by a
entity: grid no x resource set S, is in terms of readings of a canonical
quality: (entropy < 0.001)) sensor, A, as follows. Let a particular task X require, on
average, na consecutive schedules of the standard sensor A
The sensor manager uses a combinatorial auction to be completed. Instead, if a resource bundle S is used at
mechanism to find the optimal allocation given the time t = 1, the expected number of standard sensor
consumer bids. However, the consumer bids are on high- readings required is reduced to n’a. The percentage of task
level task and the sensor resources are raw sensor completed by resource set S is equal to the percentage
schedules or communication bandwidth etc. Thus, it is savings in the required number of sensor A readings.
clear that some method for bundling goods produced by
fS X,t = ( na – n’a)/ na
various sellers is essential to create commodities that
consumers are interested in and can bid for. The sensor Equation 1: Calculation of fraction of task computed
manager uses a special-purpose protocol CCA (continuous based on readings of canonical sensor
combinatorial auction) to disintegrate the high-level
consumer bids to bids for sensor resources. CCA has been Then, the bid-price for resource set S at time t for task
designed to decrease computational and communication X with bid price P is
complexity of the market. The details of CCA are ps,X,,t = fs,X,t*P
available in [10]. We briefly describe the salient features Equation 2 : Calculation of bid-price for resource set
of CCA. CCA uses discrete time slots to schedule S at time t for task X
resources. For each time slot, a combinatorial auction is
held to figure out the optimal allocation of resources.
The number of resource combinations that can be
allocated to a task is exponential in the number of
Consume resources. For large sensor networks, we have formulated
Finished Goods a special genetic algorithm that solves the bid translation
Consumers problem in polynomial time [10]. The representation
Area scans, target tracks etc
schema used for the algorithms resembles the SGA
algorithm that we developed for determining
Market
Bid combinatorial auction winners [11]. First, let the number
Reports of bids be k and the number of sensors be n. The
Budgets
representation schema is a real-valued string of length n,
Sensors where each string bit is a real-valued number between 1
and k. The jth string member essentially represents the bid
Sensor Transmission Schedule Sensor Mission
Pointing, Manager to which sensor j is allocated. We obtain this string’s
Battery Power,
Channels Manager
Market auction overall utility as the sum of utilities obtained from
Processing Bandwidth
individual bids, which are calculated as the sum of
power.
individual bid-prices calculated as shown in Equation 2.
The resulting algorithm is anytime and has polynomial
Figure 1 Architecture of MASM complexity.
References: