Publication | Closed Access
CellSlice: Cellular wireless resource slicing for active RAN sharing
131
Citations
11
References
2013
Year
Unknown Venue
Mobile Data OffloadingCross-layer OptimizationEngineeringNetwork SlicingEdge ComputingCellular Wireless ResourceCloud ComputingComputer EngineeringDownlink SlicingMulti-access Edge ComputingMobile ComputingHeterogeneous NetworkDevice-to-deviceEffective Downlink Slicing
Slicing at the gateway level is difficult because basestations make fine‑grained scheduling decisions that gateways cannot see. The paper introduces Cell‑Slice, a system that slices wireless resources to enable efficient RAN sharing. CellSlice is a gateway‑level solution that slices resources without modifying basestations’ MAC schedulers, using a feedback‑based adaptation algorithm for uplink and a technique similar to NVS for downlink, and is implemented on a Picochip WiMAX testbed. Prototype evaluation and simulations show that CellSlice achieves performance close to NVS for both uplink and downlink slicing, and its design is access‑technology independent, applicable to LTE, LTE‑Advanced, and WiMAX.
We present the design and implementation of Cell-Slice, a novel system for slicing wireless resources in a cellular network for effective Radio Access Network (RAN) sharing. CellSlice is a gateway-level solution that achieves the slicing without modifying the basestations' MAC schedulers, thereby significantly reducing the barrier for its adoption. Achieving slicing with a gateway-level solution is challenging, however, since resource scheduling decisions occur at the basestations at fine timescales, and these decisions are not visible at the gateways. In the uplink direction, CellSlice overcomes the challenge by indirectly constraining the uplink scheduler's decisions using a simple feedback-based adaptation algorithm. For downlink, we build on the technique used by NVS, a native basestation virtualization solution, and show that effective downlink slicing can be easily achieved without modifying basestation schedulers. We instantiate a prototype of CellSlice on a Picochip WiMAX testbed. Through both prototype evaluation and simulations, we demonstrate that CellSlice's performance for both remote uplink and remote downlink slicing is close to that of NVS. CellSlice's design is access-technology independent, and hence can be equally applicable to LTE, LTE-Advanced and WiMAX networks.
| Year | Citations | |
|---|---|---|
Page 1
Page 1