Journal of Systems Engineering and Electronics
Vol. 24, No. 1, February 2013, pp.165–172
Troubleshooting algorithm for solving
assignment problem and its applications
Li Zhou
∗
, Hailin Zou, Yancun Yang, and Qian Gao
School of Information and Electrical Engineering, Ludong University, Yantai 264025, China
Abstract:
A new troubleshooting algorithm for solving assign-
ment problem based on existing algorithms is proposed, and an
analysis on the related theory is given. By applying the new
troubleshooting algor ithm to the Lagrange relaxation algorithm of
the multi-dimensional assignment problem of data association for
multi-passive-sensor multi-target location systems, and comparing
the simulation results with that of the Hungarian algorithm which is
the classical optimal solving algorithm, and the multi-layer order-
searching algorithm which is a sub-optimal solving algorithm, the
performance and applying conditions of the new algorithm are
summarized. Theory analysis and simulation results prove the
effectiveness and superiority of the new algorithm.
Keyw ords: assignment problem, troubleshooting algorithm, data
association.
DOI: 10.1109 /JSEE.2013.00021
1. Introduction
The technologies of multi-sensor multi-target locating and
tracking have impor tant application s in bo th militar y and
civilian fields. In order to seize the initiative in elec-
tronic warfare, passive detection and tracking technolo-
gies are often used in detection, location, identification,
guidance, and communication systems, etc. These tech-
nologies are good for hiding and have high survivability,
strong anti-jamming capability and the anti-attack abil-
ity of anti-radiation missiles, etc. And these character-
istics are very important in improving the system’s sur-
vivability in electronic warfare [1,2] . In practical d e-
tection, the feature data obtained from the enemy are
very limited and uncertain, and usually the target’s state
measurement is the only important and reliable infor-
mation. Therefore, using azimuth information to track
the target in multi-target tracking is a very important
Manuscript received September 02, 2011.
*Corresponding author.
This work was supported by the National Natural Science Founda-
tion of China (61170161), the Natural Science Foundation of Shandong
(ZR2009GM002), and the Technology Projects of Shandong University
(J09LG01).
branch of technologies in multi-target passive tracking.
In applications, multi-passive-sensor is often used in the
direction-finding cross-location system, while the premise
of obtaining the accurate targets’ position estimate is to
associate the measurements with targets correctly. The al-
gorithms of data association of multi-target mainly include
joint probability data association algorithm [3–5], multi-
ple hypothesis algorithm [6,7] and multi-dimensional (S-
D) assignment algorithm [8–11], etc.
The algorithm of data association for a multi-sensor
multi-target system is to decompose the data received from
different sensors into different measurement sets or tracks.
Once the tracks h ave been formed or confirmed, the num-
ber of targets to be tracked as well as the state of each
target can b e estimated. The multi-sensor joint p roba-
bilistic data association algorithm estimates the target’s
state by calcu lating the association probability of measure-
ments falling into the threshold with tracks. The calcu-
lation complexity grows exponentially with the increase
of targets and effective echoes. The multiple hypothe-
ses testing (MHT) algorithm estimates the state of the tar-
get under certain criteria by enumerating the possible hy-
pothesis and association probability of the measurement
with the track in a period of time. The complexity of
the exhaustive MHT algorithm is unacceptable. We must
limit the searching times in applications. Th e S-D assign-
ment model of data association is an optimization prob-
lem with a certain number of constraints. If the S-D as-
signment problem has dimensions equal to or bigger than
three, then the complexity will grow exponentially with
the increase of the dimension of the assignment prob-
lem. Therefore the S-D assignment problem is an NP-
hard problem. References [8–10] have made a deeply
study for the sub-optimal algorithm, i.e., Lagrange re-
laxation algorithm, and have done a large number of si-
mulations. The simulation results have shown that the al-
gorithm is a polynomial-time algorithm. But the calcu-
lation burden is still heavy with the increase of the di-
mension of the cost matrix. Reference [12] expanded