Nowadays, electricity has become a basic requirement in daily life. Our world is now seeking to improve the traditional electrical grid (EG) and turn it into a smart grid (SG), due to the limitations in the current electrical networks. The traditional electrical grid (EG) is a system to deliver electricity to the consumer for daily use. It is about interconnected networks, which have power plants, generators, transmission lines, transformers, substations, distribution lines, and consumption points. Due to the advances of technology, many electricity companies all over the world are moving towards the smart grid (SG) the next generation of the electric grid for providing better reliable and cost-effective services to their customers. The smart grid (SG) is a combination of the power system, communication, information technology, and control infrastructures that can monitor, manage and optimize the operations of all parts of the power system from power generation, power transportation, and power distribution. Demand Side Management (DSM) is an energy management function of the smart grid that from the consumer's perspective is about reducing the electricity consumption during peak hours and enabling a continuous non-interruptible power supply during the peak demand hours. This thesis presents a demand-side management strategy based on a load shifting technique for demand-side management of future smart grids. A service called Load Shifter for Smart Grid Resources Optimization Middleware (SGROM-LS) is proposed in this thesis. We provide an analytical model that is formulated as a mathematical cost minimization problem. The proposed resource optimization management middleware SGROM- LS aims at reducing the consumer's bill, reducing power consumption at the peak hours (avoid power disconnection), and reducing the grid system's energy losses. The Genetic Algorithm (GA) approach is used to model the problem by collecting day ahead forecast and Service Level Agreements (SLA) from the consumers. Consumers' forecasts are represented by a two-dimensional chromosome where a gene consists of allocating a specific consumer's power task for a specific time slot (24-time slots). The output of the GA is the best power task plan for the day for all the consumers which minimizes the cost and reduces the peak demand. The experimental results show that the proposed strategy achieves substantial savings (38%) in the total cost which will be reflected in the consumers' bills and reduces the peak load demand of the smart grid to save the electrical infrastructure for an electricity company.