Abstract : The alliance between the Smart Home embedded devices e.g., gateways, tablets, media server, home automation boxes, and the Cloud for the Smart Home Service hosting becomes a proven reality. Taking into account Cloud to overcome the embedded resources limitation e.g. for reducing the bill of material for a new customer, or to offer service flexibility and scalability is a very appealing goal. In order to take benefit of these opportunities, selecting an optimal deployment of Service applications in this distributed platform is required. We introduce a new approach based on feature modelling which has been originally introduced for SPL for expressing the deployment constraints in spite of the Smart Home wide variability. As a first step we show how to get an optimized first deployment of any given set of Home Services. In a second step we will address the deployment adaptation at runtime in response to environment or software evolution.