The reference C-RAN architecture used in the chapter is shown in figure 4.1. A set of RRUs in an area is divided into groups and connected to different nodes of the transport network, called transport nodes. Transport nodes are interconnected using optical fibers, creating the so-called fronthaul network. Each transport node hosts an edge Data Center (DC) with computational resources that can be used to execute the distributed algorithm locally and to virtualize baseband resources, whenever needed. Transport nodes are all potential candidates to host BBU hotels, where several BBUs, either physical or virtualized and accessed through ports, can be deployed. All RRUs directly connected to the same transport node is assumed to be assigned to the same BBU hotel to apply interference mitigation in the considered area [51].
Each RRU has a dedicated BBU port in the primary BBU hotel and an additional port on a backup BBU hotel assigned for reliability, possibly shared with other RRUs as it will be explained later. The fronthaul network imposes strict latency and capacity demands [52], and usually requires dedicated resources to satisfy these requirements. The traditional packet-based backhaul provides connectivity between BBU hotels and the core network of mobile network providers (not reported in the figure).
The survivable BBU hotel placement problem addressed in this chapter is defined as follows:
Table 4.1: List of cost variables and parameters. Parameters:
S Set of transport nodes, |S| = s. CB The cost of activation of BBU hotel.
CH The cost of distance between each BBU-RRU pair.
CP The total cost backup and primary BBU ports.
H s × s matrix. hij is the distance in hops between nodes i and j computed
with the shortest path.
α Weight for the distance in the cost function. β Activation cost for a single BBU hotel. γ Cost of a BBU hotel port.
Variables:
Bi 1 if node i ∈ S hosts a BBU hotel, 0 otherwise.
pij 1 if BBU hotel i is assigned as primary for RRUs at node j,0 otherwise.
bij 1 if BBU hotel i is assigned as backup for RRUs at node j, 0 otherwise.
xi Number of BBU ports required at hotel site i for primary purposes.
yi Number of BBU ports required at hotel site i for backup purposes.
garding i ) total number of connected RRUs, ii ) directly connected neighbor transport nodes, iii ) maximum number of wavelengths in each link, and iv ) maximum allowed distance to connect RRUs with BBUs.
• Find: a minimum cost primary BBU hotel placement so that each RRU is assigned to a BBU according to limitations on distance and wavelength avail- ability for fronthaul links, and a minimum cost backup BBU hotel placement for reliability against single BBU hotel failure.
In the following, some useful parameters and variables are defined, while the notation used throughout this section is summarized in table 4.1.
To provide reliability against single BBU hotel failure, each RRU is connected simultaneously to two BBU hotels placed in different transport nodes, one for pri- mary and one for backup purposes. The activation cost of BBU hotels in transport nodes needed to provide full coverage and resiliency of the target area is calculated using the following formula:
CB = β ·
X
i∈S
Bi (4.1)
where Bi is a boolean variable equal to 1 when the transport node hosts a BBU
hotel, that is when it requires BBU functionalities, either for primary or backup purposes, related to one or more RRUs. β is a parameter associated with the activation cost for a BBU hotel in transport nodes and can be set accordingly.
To account for the delay introduced in the fronthaul network, a cost can be associated with the distance between BBU hotels and RRUs connecting to them. Distance between adjacent transport nodes is here assumed to be equal to 1 hop for all links. The overall cost for the distance is expressed as follows:
CH = α · X i∈S X j∈S pijhij + α · X i∈S X j∈S bijhij (4.2)
where pij and bij are boolean variables that indicate if BBU hotel i is assigned
as primary or a backup, respectively, for the group of RRUs at transport node j. hij represents the distance, in hops, between transport node i and j computed
solving the shortest path problem. Both contributions (i.e., the overall distance for the primary path and backup path) are multiplied by cost parameter α, which represents the cost for the 1 hop link.
Finally, the proper number of BBU ports must be allocated in each hotel. The total number of primary and backup BBU ports, and the related cost, are calculated according to the following formula:
CP = γ ·
X
i∈S
xi+ yi (4.3)
CP is the contribution of the total number of primary xi and backup yi ports
in each hotel multiplied by the cost parameter γ associated with each port. Since the protection requires that each RRU is connected to two different BBU hotels, the total number of ports should be twice the number of RRUs, and consequently, the value for CP can be fixed. However, only the number of primary ports is fixed
and equal to the number of RRUs. On the contrary, the number of backup ports can be reduced. RRUs can share backup ports if they have different primary BBU hotels. When a single hotel failure occurs, RRUs assigned to that primary hotel switch to their backup hotel, hence it is forbidden to share backup ports among RRUs assigned to the same primary. By sharing the backup ports, the value for CP
can be reduced, and further cost saving can be achieved.