EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2161. Large Scale Topology and Connectivity Design with In-Network Processing
Invited abstract in session TD-57: Large-scale network optimization and inventory management, stream Optimization at Amazon.
Tuesday, 14:30-16:00Room: S06 (building: 101)
Authors (first author is the speaker)
| 1. | . Faheem
|
| SNT, Amazon | |
| 2. | Onur Can Saka
|
| SNT, Amazon | |
| 3. | George Iosifidis
|
| SNT, Amazon | |
| 4. | Amit Kumar
|
| SNT, Amazon |
Abstract
Topology and connectivity design with in-network processing (sortation, cross docking) is a critical component for Middle Mile transportation network design to achieve operational efficiency, cost-effectiveness, and delivery speed. In this paper, we consider a middle mile transportation network that consists of warehouses, sortation centers and delivery stations. Given a warehouse to delivery station volume forecast, our goal is to design cost and speed optimal transportation network. We formulate the topology and connectivity design problem as a Mixed Integer Linear Programming (MILP) problem, which is NP-hard. We describe the key characteristics of the problem and analyze the arising modeling and optimization challenges. To solve large scale real-world problems (>100 million variables), we present our optimization-based matheuristic pipeline, which comprises preprocessing and path pruning strategies, local neighborhood search, and incremental model solving. We show the efficacy of our model using a range of operational datasets. We conclude with potential research threads that can further improve the solution quality and runtime.
Keywords
- Network Design
- Transportation
Status: accepted
Back to the list of papers